Gtx 580 vs gtx 480: GeForce GTX 580 vs 480 [in 5 benchmarks]

AMD Radeon 535

Nvidia GeForce GTX 480

vs

AMD Radeon RX 480

Nvidia GeForce GTX 580

vs

Nvidia GeForce GT 1030 DDR4

Nvidia GeForce GTX 480

vs

Nvidia GeForce GTX 750 Ti

Nvidia GeForce GTX 580

vs

AMD Radeon RX 550

Nvidia GeForce GTX 480

vs

Zotac GeForce GT 240 AMP! Edition

Nvidia GeForce GTX 580

VS

NVIDIA GEFORCE GTX 660 Ti

Complexation of prices

Users reviews

Productivity

1. TECTION FECTION GP

701MHz

772MHZ

GPU GPU (GPU) has a higher so -so -so -so -so -so -so -so -so -so -so -so -so -so -so -free frequency frequency frequency frequency.

2.turbo GPU

Unknown. Help us offer a price. (Nvidia GeForce GTX 480)

Unknown. Help us offer a price. (Nvidia GeForce GTX 580)

When the GPU is running below its limits, it may jump to a higher clock speed to increase performance.

3.pixel rate

21 GPixel/s

24.7 GPixel/s

The number of pixels that can be displayed on the screen every second.

4.flops

1.34 TFLOPS

1.58 TFLOPS

FLOPS is a measure of GPU processing power.

5. texture size

42.1 GTexels/s

49.4 GTexels/s

Number of textured pixels that can be displayed on the screen every second.

6. GPU memory speed

924MHz

1002MHz

Memory speed is one aspect that determines memory bandwidth.

7.shading patterns

Shading units (or stream processors) are small processors in a video card that are responsible for processing various aspects of an image.

8.textured units (TMUs)

TMUs accept textured units and bind them to the geometric layout of the 3D scene. More TMUs generally means texture information is processed faster.

9 ROPs

ROPs are responsible for some of the final steps of the rendering process, such as writing the final pixel data to memory and for performing other tasks such as anti-aliasing to improve the appearance of graphics.

Memory

1.memory effective speed

3696MHz

4008MHz

The effective memory clock frequency is calculated from the memory size and data transfer rate. A higher clock speed can give better performance in games and other applications.

2.max memory bandwidth

177.4GB/s

192.4GB/s

This is the maximum rate at which data can be read from or stored in memory.

3.VRAM

VRAM (video RAM) is the dedicated memory of the graphics card. More VRAM usually allows you to run games at higher settings, especially for things like texture resolution.

4.memory bus width

384bit

384bit

Wider memory bus means it can carry more data per cycle. This is an important factor in memory performance, and therefore the overall performance of the graphics card.

5.versions of GDDR memory

Later versions of GDDR memory offer improvements such as higher data transfer rates, which improve performance.

6. Supports memory debug code

✖Nvidia GeForce GTX 480

✖Nvidia GeForce GTX 580

Memory debug code can detect and fix data corruption. It is used when necessary to avoid distortion, such as in scientific computing or when starting a server.

Functions

1.DirectX version

DirectX is used in games with a new version that supports better graphics.

2nd version of OpenGL

The newer version of OpenGL, the better graphics quality in games.

OpenCL version 3.

Some applications use OpenCL to use the power of the graphics processing unit (GPU) for non-graphical computing. Newer versions are more functional and better quality.

4. Supports multi-monitor technology

✔Nvidia GeForce GTX 480

✔Nvidia GeForce GTX 580

The video card has the ability to connect multiple screens. This allows you to set up multiple monitors at the same time to create a more immersive gaming experience, such as a wider field of view.

5. GPU temperature at boot

Unknown. Help us offer a price. (Nvidia GeForce GTX 480)

Lower boot temperature means the card generates less heat and the cooling system works better.

6.supports ray tracing

✖Nvidia GeForce GTX 480

✖Nvidia GeForce GTX 580

Ray tracing is an advanced light rendering technique that provides more realistic lighting, shadows and reflections in games.

7. Supports 3D

✔Nvidia GeForce GTX 480

✔Nvidia GeForce GTX 580

Allows you to view in 3D (if you have a 3D screen and glasses).

8.supports DLSS

✖Nvidia GeForce GTX 480

✖Nvidia GeForce GTX 580

DLSS (Deep Learning Super Sampling) is an AI based scaling technology. This allows the graphics card to render games at lower resolutions and upscale them to higher resolutions with near-native visual quality and improved performance. DLSS is only available in some games.

9. PassMark result (G3D)

This test measures the graphics performance of a graphics card. Source: Pass Mark.

Ports

1.has HDMI output

✔Nvidia GeForce GTX 480

✔Nvidia GeForce GTX 580

Devices with HDMI or mini HDMI ports can stream HD video and audio to an attached display.

2.HDMI connectors

Unknown. Help us offer a price. (Nvidia GeForce GTX 480)

Unknown. Help us offer a price. (Nvidia GeForce GTX 580)

More HDMI connectors allow you to connect multiple devices such as game consoles and TVs at the same time.

HDMI version 3

Unknown. Help us offer a price. (Nvidia GeForce GTX 480)

Unknown. Help us offer a price. (Nvidia GeForce GTX 580)

Newer versions of HDMI support higher bandwidth for higher resolutions and frame rates.

4. DisplayPort outputs

Allows connection to a display using DisplayPort.

5.DVI outputs

Allows connection to a display using DVI.

6.mini DisplayPort outputs

Allows connection to a display using mini DisplayPort.

Price Match

Cancel

Which graphics cards are better?

gtx 480 | Tags | Overclockers.ru

golds

July 10, 2021

If this happens, it will be the first card in almost 10 years related to this

series

ddr22

October 24, 2018

12 flagship benchmarks: 8800 GTX, 8800 GT, 9800 GTX+, GTX 280, GTX 480, GTX 580, GTX 680, GTX 780 Ti, GTX 980, GTX 980 Ti, GTX 1080, GTX 1080 Ti

Dmitry Vladimirovich

October 19, 2015

A lot of copies are broken due to old video cards and new drivers. And there is still no clarity on this issue. We decided to conduct a little research and for this purpose took the GeForce GTX 480 — the flagship of 2010, and several subsequent top models — the GTX 580, GTX 680 and GTX 780. Another no less relevant question is after how many years or generations of video cards does it make sense to upgrade ?

Phoenix

October 12, 2014

Testing twenty-four NVIDIA GeForce GTX 4xx, GTX 5xx, GTX 6xx and GTX 7xx video cards in ten games, 1920×1080 resolution and two operating modes.

Phoenix

September 7, 2014

Twelve graphics cards tested in ten games, 1920 x 1080 resolution, and four operating modes.

Phoenix

June 1, 2014

Testing four processors and eight video cards in twelve games, one resolution and two operating modes.

Phoenix

May 25, 2014

Testing four processors and eight video cards in twelve games, one resolution and two operating modes.

Phoenix

April 2, 2014

Testing twenty-five video cards in twelve games, two operating modes and resolutions of 1920 x 1080 and 2560 x 1440.

Dmitry Vladimirovich

February 25, 2014

Not even a day has passed since the announcement of the older model based on Maxwell, when video cards went on sale. This time, using the ASUS GeForce GTX 750 (GTX750-PHOC-1GD5) as an example, we will consider the younger solution — GTX 750. Without thinking twice, NVIDIA decided not to change the frequency characteristics of the new graphics processor, but to truncate the number of PS and texture units by exactly one GPC.

Dmitry Vladimirovich

February 18, 2014

Official announcement of NVIDIA GeForce GTX 750 Ti graphics accelerator. Maxwell architecture and new model PCB features, overclocking, benchmarking with Radeon R7 240, R7 250, R7 260X, R7 265, GeForce GTX 480 and GTX 650 Ti Boost.

Dmitry Vladimirovich

February 10, 2014

Let’s check how expedient it is to increase the speed of the graphics subsystem by adding a discrete video card to the Kaveri APU. For clarity, we will conduct tests both in joint mode and separately. The participants included the A10-7850K, A8-7600, Radeon R7 250 and R7 240. Let’s add to them the freestanding R7 260X, and for a change, we’ll take the top model of four years ago GeForce GTX 480.

Phoenix

July 2, 2013

Summary testing of three generations of video cards in two resolutions and two operating modes.

Phoenix

May 12, 2013

Reference information for AMD and Intel processors, and AMD and NVIDIA graphics cards.

Phoenix

September 5, 2012

Author: Phoenix. Testing processors Intel Core i7-3770K, Core i5-3570K, Core i7-2600K, Core i5-2500K, Core i7-930, Core i7-860, Core i5-760, Core 2 Quad Q9500 in sixteen games, three resolutions and four operating modes.

Phoenix

August 29, 2012

Author: Phoenix. Testing thirty-five video cards in sixteen games, four resolutions and two operating modes.

Phoenix

August 16, 2012

Author: Phoenix. Testing twenty-eight processors in sixteen games, three resolutions and two operating modes.

Phoenix

July 30, 2012

Author: Phoenix. AMD Radeon HD 7xx0, HD 6xx0, HD 5xx0 and NVIDIA GeForce GTX 6×0, GTX 5×0, GTX 4×0 against Crytek projects.

Phoenix

July 24, 2012

Author: Phoenix. Testing processors Core i7-2600K, Core i5-2500K, Core i3-2120, Core i7-930, Core i7-860, Core i5-660, Pentium G6960 V sixteen games and three resolutions.

Phoenix

May 5, 2012

Author: Phoenix. Testing of processors Pentium G850, Pentium G840, Pentium G630, Pentium G620, Celeron G540, Celeron G530, Pentium G6960, Athlon II X3 455, Athlon II X3 440, Athlon II X2 265, Athlon II X2 250 and Athlon II X2 240, in sixteen games, three resolutions and two modes of operation.

Phoenix

April 26, 2012

Author: Phoenix. Testing processors FX-8150 BE, FX-8120 BE, FX-6100 BE, Core i7-2600K, Core i5-2500K, Core i5-2300, Core i7-860, Core i5-760, Phenom II X6 1100T BE, Phenom II X4 965 BE, Phenom II X4 850 and Athlon II X4 640 in sixteen games, three resolutions and two operating modes.

Phoenix

March 24, 2012

Author: Phoenix. Testing twenty-four video cards in sixteen games, four resolutions and two operating modes.

Phoenix

March 3, 2012

Author: Phoenix. Testing processors FX-4100, Core i3-2120, Pentium G860, Pentium G630, Core i5-660, Core i3-560, Pentium G6960, Phenom II X4 965 BE, Phenom II X4 850, Phenom II X2 565 BE, Athlon II X4 640, Athlon II X3 440, Athlon II X2 265 in sixteen games, three resolutions and two operating modes.

Phoenix

February 20, 2012

Author: Phoenix. Testing thirty-eight processors in sixteen games, three resolutions and two operating modes.

Dmitry Vladimirovich

December 2, 2011

What happens if the line of GeForce GTX 560 video cards based on GF114 is expanded by adding a graphics accelerator with the GF110 core?

Phoenix

August 7, 2011

Author: Phoenix

Phoenix

July 17, 2011

Author: Phoenix

[Viru$]

July 7, 2011

We install the following processor coolers on the GeForce GTX 550 Ti GS, GTX 460 and GTX 480: Scythe Samurai ZZ, Zalman CPNS-7000AlCu and a boxed cooler from the Intel Q6600 processor.

Phoenix

July 3, 2011

Author: Phoenix

Phoenix

June 28, 2011

Author: Phoenix

Dmitry Vladimirovich

June 17, 2011

Overclocking MSI GTX 580 Lightning with freon and comparison of two generations of Lightnings — GTX 480 and GTX 580. Testing in synthetic and gaming applications.

[Viru$]

June 10, 2011

We continue our practical exercises. MSI AfterBurner, soft overclocking and stability testing with Inno3D GeForce GTX 480 iChiLL, Inno3d GeForce GTX 460, Gainward GTX 550 Ti GS, AMD Radeon HD 6870 and AMD Radeon HD 5850.

Phoenix

May 22, 2011

Author: Phoenix

Phoenix

May 15, 2011

Author: Phoenix

Phoenix

May 8, 2011

Author: Phoenix

Phoenix

May 1, 2011

Author: Phoenix

Dmitry Vladimirovich

March 24, 2011

Testing the new NVIDIA flagship — GeForce GTX 590. Comparison with GTX 480, GTX 580, GTX 590, HD 6990.

Dmitry Vladimirovich

March 8, 2011

The Radeon HD 6990 is AMD’s new dual processor graphics accelerator. Review and comparison with HD 5970, HD 6970, HD 6950, HD 6870, HD 6850, GTX 580, GTX 570, GTX 560 Ti, GTX 480, GTX 470, GTX 465, GTX 460.

wildchaser

March 3, 2011

Two «frostbitten» graphics accelerators based on video processors from different manufacturers. Review, overclocking, comparison with Gainward GTX570 GLH and Zotac GeForce GTX 480 AMP!.

Review and test NVIDIA GeForce GTX 580

Released in March, the NVIDIA GeForce GTX 580 graphics card has been described as the fastest, hottest and most expensive. But it doesn’t make much sense to compete with the ATI Radeon HD 5870, yes, the main advantage of the video card is its speed (15-20%), but the high price does not justify the difference. There are a number of reasons that influenced the Fermi architecture. One of the most important is the development of an architecture aimed at professional use for complex computing, the increased complexity, entailed an increased number of transistors in the GPU, which led to an increase in it, and this contributed to certain production difficulties and increased cost, high power consumption and heat dissipation. But if we consider the model GeForce GTX 480 from the point of view of marketing, then it has definitely achieved high consumer ratings, and is considered the fastest single-chip video card.

Today, the manufacturer is releasing a new GTX 580, the successor to the previous model, which promises to fix the success and eliminate some of the shortcomings. At the first inspection of the novelty, it is already possible to understand that the new released model is a modified old generation of graphics. But this is not a reproach, for example, AMD did exactly the same when it released, under the guise of new developments, upgraded and cheaper models of Radeon HD 6800 video cards. Also, for many current games, the performance of the video card is enough, so the performance increase is relevant only for the highest resolutions and 3D Vision technologies that increase the load on the GPU. According to experts, in the future, the number of gaming projects, especially resource-intensive ones, will be reduced, and the most basic of them will be developed with an emphasis on consoles, where the performance is very inferior to a PC.

Architecture

The new GeForce GTX 580 is based on the GF110 GPU which contains 3 billion transistors. It should be noted that the technological lines of the semiconductor company TSMC, which produces graphics processors for NVIDIA and AMD, have been significantly improved, in terms of power consumption and heat dissipation, the new chip will be improved.

The GF110 processor combines the unambiguous structure of the predecessor model and the updates that have been achieved throughout the year. Unlike the previous model from the GeForce GTX 480, all blocks are active in the updated processor, they have 512 flexible and efficient shader processors, 64 texture units and 16 Polymoph Engine units. All other characteristics remained untouched, these are 48 rasterization units, and a 768 KB L2 cache, and 64-bit memory controllers that form a 384-bit bus, and the novelty is also equipped with 1.5 GB of memory. Based on all of the above, it seems that the new GTX 580 graphics card is not that far removed from the former leader, but there are other innovations. One of them is increased clock speeds, the current GPU runs at 772 MHz, and the shredder domain at 1544 MHz. Memory increased to 1 GHz.

The second and most important innovation includes updated texture units, which first appeared on the GTX 460. Their difference lies in twice the speed increase, while working with FP16 textures, which are used in modern games, which means that HDR mode is processed at high speed and does not lose performance. And finally, another important improvement is a redesigned pixel culling algorithm (Z-cull), with which you can more accurately determine the hidden parts of the scene, extract them from processing, while freeing up resources. The manufacturer claims that the increase in this operation is 8%, which makes a significant contribution to productivity.

Troubleshooting

Naturally, the manufacturer paid special attention to Fermi features that had complaints, ie energy consumption, heat dissipation and noise. Absolutely all GF110 functional blocks are redesigned, taking into account all the flaws that were found in the GF100 blocks, and now the GPU has less current leakage, and therefore it heats up less. In the process of modernization, the power supply system has also changed, in terms of both the element base and management. Now it includes three sensors that measure the current consumed by the video card. If the driver notices an abnormal increase in current and the use of utilities, then the clock speeds are reduced to maintain the declared TDP level.

Of course, it’s not good to compare with competitors, but NVIDIA can understand that very high power consumption rates are determined using utilities, and, as you know, many do not run them, and the load on the video card in modern games is not found at all. Today, we can observe similar situations with CPU thermal packs, in the understanding of manufacturers AMD and Intel, AMD applies the average consumption indicator (ACP), and Intel applies the maximum allowable TDP. However, it should be noted that this algorithm is used only in the above applications.

As for the cooling system, it has been significantly changed, so instead of heat pipes, the radiator has a large chamber, which is located over the entire area of ​​its sole, and thin aluminum fins. The already familiar centrifugal fan blows hot air out. The cooler control and thermal control systems have also been upgraded. The characteristic sharp jump in the fan speed and its noise is reduced due to a smooth increase. As a result, at maximum load on the video card, the noise is clearly audible, but this does not interfere with comfortable work. It is also worth considering that the temperature regime has improved significantly. GeForce GTX 580, even at the highest load with the frequency reduction algorithm turned off, does not even reach 90 °.

Test results

While testing the GTX 580, you can see a bunch of minor improvements. Which led to a significant speed of the video card. The superiority of the new model lies in higher performance in texturing (11-28%), faster tessellation and increased frequencies. If earlier it was said that the GTX 480 is the fastest single-chip video card, and the dual-processor Radeon HD 5970 video card is significantly ahead of it, now we can safely say that the new product from NVIDIA occupies all the leading positions. In certain tests, the AMD graphics card outperforms the GTX 580, only in 3DMark Vantage and Aliens vs. Predator, and in the rest, the new Fermi, with a minimal margin ahead of the HD 59 graphics card70, and given that only the first version of the driver was released, we can assume that in the future, the gap will increase significantly. It should be noted that in heavy games, power consumption has significantly decreased, these figures have decreased by 30-35 watts. During the heaviest load of the FurMark boost mode, with the software current limit turned off, the GTX 580 has the same performance, similar to the previous version, and this attracts attention with the increase in functional blocks and the frequency of optimization of the core and power system.

Conclusions and energy consumption is significantly reduced, as well as it is really fast. Of course, it’s too early to talk about the overthrow of the Radeon HD 5970 video card from its pedestal, but its performance and the performance of the new product are almost on a par, and if we also take into account ease of use, then the GeForce GTX 580 clearly occupies the highest positions. So far, it makes no sense to compare the new products from NVIDIA and AMD, although video cards have a difference in price, but they also have a greater difference in performance.

NVIDIA GeForce GTX 580 Pricing News — WORLD NVIDIA

GeForce GTX 580 price tags have already appeared online. , PNY, MSI, EVGA.

So, the price tag for the ASUS GeForce GTX 580 has already slipped on Amazon — ASUS ENGTX580 / 2DI / 1536MD5 cost $ 519 there, however, after the network media publicized this case, the information was quickly covered up.

However, a holy place is never empty — and now one of the American retailers has published the price of the GeForce GTX 580 from EVGA. The price is divine — only $ 499, shipping — on the 9th (i.e. tonight can be expected).

Canadians are not far behind the Americans — their retailers are already selling PNY GeForce GTX 580 for 566 and GeForce MSI GTX 580 for 580 Canadian dollars.

So it looks like the ice has broken…

rumorsvideo cardsEVGANVIDIAFermiGeForce GTX 580

comment related news

Fudzilla

November 5, 2010

Comparison with Radeon HD 5870, 6870 and GTX 480 plus SLI vs Crossfire.

Information about the upcoming NVIDIA flagship GeForce GTX 580 continues to circulate on the network. This time, one of the Italian IT resources gave birth to slides allegedly from NVIDIA’s presentation containing «fat-free» results of comparative benchmarks of the GTX 580 with other video cards, both from NVIDIA and AMD.

The other day we have already published data on the performance of the new GeForce, so those who follow the news will be interested to see the «second series»:

the load will be lower than that of the GTX 480. It is even more difficult to believe in this than in such a total lead over competitors’ solutions.

rumorsvideo cardsNVIDIAFermiGeForce GTX 580benchmarks

comment on similar news

News4it

November 3, 2010

Faster than the GeForce GTX 480, but not fast enough to compete with future AMD solutions.

The Chinese site PCOnline.com managed to get the performance results of the upcoming GeForce GTX 580 in comparison with the GeForce GTX 480 and AMD Radeon HD 5870. Testing was carried out in 18 games and benchmarks.

The GeForce GTX 580 showed in some places (in the DirectX 11 Stone Giant and Heaven benchmarks) more than a twofold advantage over the Radeon HD 5870. The performance increase compared to the GTX 480 in most cases was no more than 15%. However, in DX9/10 applications GTX 580 is unlikely to be able to keep the crown of the leader after the release of solutions AMD Radeon 6990/6970.

Some details have also appeared about other characteristics of the GTX 580 video cards. Information about the presence of 128 TMUs in the chip has been added to the already known information, which again raises the question of which chip the cards will be based on, GF100 or GF110, since the GF100 in the complete set has only 64 texture block. The “official” TDP has also become known — 244 W, which is even less than 250 W for the GTX 480. However, in applications like FurMark, the card will still “knock out” around 300 W. The date of the official announcement is determined by various astrologers and seers in the region of 8-9november.

It should be remembered that all such information before the appearance of sample maps by reviewers of trustworthy resources should be taken with skepticism.

PerekvidocardiafermigeForce GTX 580 Benchmarks

Comment similar news

VR-Zone

November 2, 2010

In addition to the specifications, the newly release date from NVIDIA-GTX 580.

. Previous ones are the previous ones there. the confirmation. It’s about:

• GPU frequency 772 MHz;
• shader frequency 1544 MHz;
• GDDR5 memory frequency 2004 MHz;
• video memory size 1536 (according to other sources — 1544) Mb;
• memory interface width 384 bits;
• memory bandwidth 192.4 Gb/s;
• 512 CUDA cores;
• TPD board 244W.

In addition, many sources claim that journalists have begun to receive map samples and the exact release date of the map is called Tuesday, 9-th of November. If the latter is true, then in a week we will already see the «real» GF100, which will withstand the expansion of the Radeon HD 6970 from AMD.

rumorsvideo cardsNVIDIAFermiGeForce GTX 580GF100b

comment on similar news

This is reportedly the GeForce RTX 30, and at least two models will be available for purchase in the coming weeks. These will be mid-range graphics cards, the RTX 3060 with 8 GB of VRAM, which is 4 GB less than the original model. The current model sells for $350, so the 8GB VRAM version will cost around $300.

NVIDIA GeForce RTX 3060

In addition, NVIDIA is also preparing an RTX 3060 Ti with 19Gb/s GDDR6X memory, similar to the RTX 3070 Ti. This card currently has 448 GB/s of memory bandwidth, while the upgraded card has 608 GB/s, 36% faster.

As for the RTX 3070 Ti with GA102, there are no confirmations on it yet, however, such models are often released in local markets, for example, exclusively for China.

Rumors Video Memory Video Cards NVIDIA GeForce RTX 3060 Ti3060

Comment ​Related news

Videocardz

14 September

According to recent rumors, NVIDIA has completed the development of the specification for the RTX 4070 video card, while the company decided to create two versions of this video card at once.

One of them will have 12 GB of GDDR6X video memory and 7680 CUDA cores, and the other one will have 10 GB and 7168 CUDA cores. In both cases, the throughput will be 21 Gb / s. The older version of the accelerators received the number PG141-SKU340/341. In addition to a 20% increase in video memory, this version also has a large thermal package — 285 W versus 250 W for the younger version PG141-SKU336/337.

NVIDIA GeForce RTX

In terms of performance, it will obviously be different. The older version will offer a speed of about 11,000 points in 3DMARK Time Spy Extreme, while the younger version will offer about 10,000, but in any case, this performance is comparable to the RTX 3090.

NVIDIA plans to introduce its new video cards in September during the GTC , which means a delay in the release of the RTX 40 series, caused by large stocks of RTX 30 in warehouses. Some observers have noted that NVIDIA may start selling the RTX 40 series in the fourth quarter of this year. The top models will be available first.

RumorsGeForce RTX 4070Video CardsNVIDIA

Comment on ​Related News

Obviously, the RTX 400 series of cards is still in development, and therefore their specifications continue to change. This time there are rumors that the upcoming RTX 4080 graphics card will receive GDDR6X memory with a bandwidth of 23 Gb / s, instead of the expected 21 Gb / s. At the same time, it also reports a reduction in power consumption from 420 W to 340 W. This information was spread by a well-known insider @kopite7kimi.

The rest of the specifications remain unchanged, namely the AD103 GPU, 9728 CUDA cores and 16 GB of 256-bit VRAM.

The RTX 4090 is also rumored to feature 23Gb/s memory, but a lot could change before release.

rumorsGeForce RTX 4080video cardsNVIDIA

comment on related news

Well known insider kopite7kimi has revealed that the upcoming RTX 4080 will have 9728 CUDA cores, down from the 10240 CUDA cores originally expected. The card is also said to have a TDP of 420W and be based on the AD103-300-A1 GPU. All this will be assembled on a PG136/139-SKU360 board with 16 GB of GDDR6X memory.

NVIDIA RTX

graphics card Thus, the number of CUDA cores will be reduced by 5%. This is unlikely to have a noticeable impact on performance, and the RTX 4080 is still expected to score over 15,000 points in TimeSpy. 9000 points. This is 66% more than the RTX 3090 Ti and 82% more than the RTX 3090 compared to the top Ampere cards. In other tests and games, the result may be different. However, this is a huge performance boost at 4K resolution.

When it comes to ray tracing performance, this number remains the biggest mystery. The card is known to feature 16,384 CUDA cores, 52% more than the RTX 3090 Ti. At the same time, the increase will be provided not only by an increase in the number of cores, but also by an increase in frequency and power consumption.

GPU-based cards codenamed Ada are expected to arrive in the fourth quarter of this year. However, according to the latest rumors, this year we will see only the RTX 4090 model.

rumors as well as some features of their specifications.

So, it is noted that the RTX 4090 and RTX 4080 cards will have a lot more in common than previously expected. Although they will be based on different GPUs: AD102 and AD103, the circuit board will be the same — PG139. Only their design versions will differ: 330 for the RTX 4090 and 360 for the RTX 4080. It is not yet clear whether the GPUs will be compatible by pins, but it is already obvious that the boards will be the same.

GeForce RTX 40

NVIDIA has not announced RTX 4080 and RTX 4070 specifications at this time, for two reasons. Firstly, they are based on different GPUs, and secondly, the specifications have not yet been approved.

In terms of timing, NVIDIA GeForce RTX 4090 is now expected in August, RTX 4080 in September, and RTX 4070 in October. However, these are not end dates. Manufacturing partners have a ton of RTX 30-series cards in stock, and obviously the last thing NVIDIA needs in these circumstances is to start shipping the next generation.

rumorsGeForce RTX 40

0video cardsNVIDIA

comment on related news

Videocardz

May 19

Well-known insider Kopite7kimi has published fresh information that a new generation of video cards will be released in mid-July.

NVIDIA’s flagship GeForce RTX 4090 accelerator will contain 126 streaming processors, which means 16128 CUDA cores. This is noticeably less than the previously expected 140-142 multiprocessors. At the same time, it is known that the AD102 chip contains 144 streaming multiprocessors, that is, 2304 cores less than possible. It is possible that the maximum configuration will be available in the RTX 4090 Ti.

In addition, the insider talked about the heat dissipation of the RTX 4090. According to him, now we are talking not about 600 W, but only about 450 W TDP. Hotter should also be the top RTX 4090 Ti. For comparison, the RTX 3090 Ti has a TDP of 450W. In terms of performance, it will be twice as fast as the RTX 3090, which consumes 350W.

In addition, Kopite reiterated its earlier claims that the RTX 4090 will have 24GB of GDDR6X video memory at 21Gb/s, which would mean 1TB/s of bandwidth on a 384-bit bus.

The last statement made by an insider concerns the timing of the release of a new generation of NVIDIA graphics cards. He reports that the GeForce RTX 40 will appear in mid-July. At the same time, RTX 409 performance solutions will be the first to enter the market.0 Ti, i.e. 21 Gb/s.

NVIDIA RXT Suprim package

This will also use the modification of GPU AD102-300. As a result, power consumption will be increased to 600 watts, which is significantly more than 450-500 watts consumed by the current flagship RTX 3090 Ti.

It will be very interesting to know the performance of this accelerator, because changes in the architecture, along with an increase in power consumption, should give a sharp jump in speed.

Rumors Lovelace Video Memory Video Cards NVIDIA

comment on related news

TweakTown

17 Feb

At CES 2022, NVIDIA unveiled the new GeForce RTX 3090 Ti, a card that is faster than the RTX 3090, the flagship of the current Ampere line.

The company promised that more information about the new product would be published at the end of the month, that is, January. But so far nothing has been heard about the RTX 3090 Ti.

When asked by The Verge, NVIDIA spokeswoman Jen Andersson reported: «We don’t have any more information on the RTX 3090 Ti at this time, but we’ll be in touch when it becomes available» .

Video card NVIDIA GeForce RTX 3090 Ti

The answer is frankly weak. The standard reply often used by press secretaries to make up for inaction. All this suggests that the RTX 3090 Ti is a failure.

In January, shortly after CES, NVIDIA allegedly asked card manufacturers to stop producing the RTX 3090 Ti. A blogger and insider on the Moore’s Law Is Dead (MLID) channel also reported that according to his data, this map is «postponed indefinitely» . He claims that the reason for this is problems with the circuit board. The power consumption of the card is 450 W, and the creation of such a card turned out to be a difficult task.

rumorsGeForce RTX 3090 Ti graphics cardsNVIDIA

comment on related news

Neowin

February 12

Turing based NVIDIA GeForce MX550 graphics card designed for thin multimedia laptops. This is a very interesting segment as it delivers comparable performance to CPU-integrated solutions, raising the question of the need for a discrete graphics card at all.

The first test of the MX550 has appeared on the PassMark website. This card scored 5014 in the G3D Mark test, which is almost identical to the Vega 8 GPU built into the AMD Ryzen 9 5900HS.

The graphics in the AMD Ryzen 9 5900HS in the same benchmark show 4968 points, which is only 0.9% less, and more than fits into the statistical error. Of course, PassMark is not the most popular graphic benchmark, and the number of results in it is very limited. Only 9 iGPU 59 results are offered for comparison00HS.

NVIDIA MX550 PassMark Test ResultsMX550 vs. Ryzen 9 5900HS PassMark

The MX550 graphics card is based on the TU117 GPU, the slowest variant of the Turing processor. It doesn’t have ray tracing or DLSS, but I don’t think anyone would want to use them given its baseline performance. So far, NVIDIA hasn’t confirmed the heat dissipation of the MX550. Most likely, we will receive more information by the time of its release, which is scheduled for «this spring».

testingrumorsVega 8Ryzen 9GPUMX550PassMarkvideocardslaptopsTuringNVIDIAbenchmarks

Videocardz

January 27

The specs of the new NVIDIA GeForce RTX 3090 Ti graphics card, which, as you are sure, will receive incredible bandwidth / 1 TB of video memory, have appeared on the Web

The RTX 3090 Ti is reported to have 24GB of GDDR6X memory at 21Gb/s. Thus, with a 384-bit bus, the total throughput will reach 1008 GB / s.

PCB GeForce RTX 3090 Ti

As for the GPU, it will contain 84 SM. In total, it will offer 10752 CUDA cores (versus 10496 cores at 82 SM in the RTX 3090). Alongside them, the card will offer next-generation RT cores, Tensor cores, and a brand new streaming multiprocessing box. The base frequency of the GPU will be 1560 MHz, and in Boost mode it will increase to 1860 MHz. The card’s heat dissipation is rated at 450W, 100W more than the RTX 3090.

PCIe 5.0 Auxiliary Power Slot

In addition, the new GeForce RTX 3090 Ti will be the first PCIe Gen 5.0 compatible graphics card with a single 6-pin auxiliary power connector that can deliver up to 600W.

rumorsGeForce RTX 3090 Ti graphics cardsNVIDIA

comment on similar news

According to rumors, the release of the video card will take place in January, but some information about it has already begun to appear. The GeForce RTX 3080 Ti Mobile graphics card will offer the highest performance in the Ampere family for mobile platforms. To do this, it will use faster memory than the RTX 3080, and increased heat dissipation. Thus, the memory speed will increase from 14 Gb / s to 16 Gb / s, and the heat dissipation will be increased from 165 watts for the current model to 175 watts.

NVIDIA RTX Mobility

The Ti version is rumored to have more CUDA and TMU cores, but these specifications are unknown. Instead of 6144 CUDA cores in the RTX 3080, the Ti version is expected to get 7424 cores.

rumor RTX 3080 Ti Mobilevideo -Cardynovidia

Comment similar news

TechPowerUp

December 18, 200616

All know that the NVIDIA is preparing a dimensional GEFORCE RTX 3050 for January, but it will be in two ways. video memory and based on different GPUs.

According to recent rumors, the RTX 3050 will be released based on the GA106-150 with 2560 CUDA cores and 8GB of VRAM, and based on the GA106-140 with 2304 CUDA cores and 4GB of VRAM. This is contrary to previous rumors that the GPU will contain 3072 CUDA cores.

NVIDIA GeForce RTX 3050

At the moment I’m wondering why the GA106 processor was required to make this board, because the GA107 almost completely corresponds to the required configuration (2560 cores, 128 bit bus). The answer may lie in the need to use lower-quality processors of a higher level, or the company’s desire to later release a version of the RTX 3050 Ti. In the second case, the use of the same GPU on different models will reduce the labor costs for the production of video cards from production partners.

Everything will be known for sure on January 4, when NVIDIA should officially present the video card. The release of the GeForce RTX 3050 should take place on January 27th.

Perekvideo -CardynvidiageForce RTX 3050

Comment similar news

VideoCardz

December 1, 2021

As you know, NVIDIA is preparing its RTX 30 video cards. Currently, the cheapest model for desktop systems is RTX 30660 , but according to rumors, NVIDIA will soon expand the lineup by adding the RTX 3050 to it.

The RTX 3050 and RTX 3050 Ti are currently only available for laptops, making this a desktop model for the first time.

NVIDIA GeForce RTX 3050

The RTX 3050 will reportedly be based on the GA106-150 GPU with 3072 CUDA cores. The memory subsystem will be represented by 8 GB GDDR6 with a bus width of 128 bits. The desktop NVIDIA RTX 3050 is said to be faster than the GTX 1660 Super, but slower than the RTX 2060 12GB. Given these estimates, we can say that the new NVIDIA accelerator is designed to strengthen the competitive position against the upcoming AMD RX 6500 XT and Intel Arc Alchemist 128EU cards.

rumorsNVIDIAGeForce RTX 3050

video cards comment on similar news This time there were rumors that the company is preparing to re-release the GeForce RTX 2060 video card next year.

NVIDIA already resumed the release of the RTX 2060 video card in January. which will receive exactly the same specifications as the existing model, with the exception of the amount of video memory, which will be 12 GB.

NVIDIA GeForce RTX 2060

The firm has already begun informing its manufacturing partners about this card, which should go on sale by the end of this year. This will give manufacturers enough time to launch the new model by 2022.

Clearly, NVIDIA’s decision to relaunch the RTX 2060 with more memory is due to the catastrophic shortage of the RTX 3060, for which speculators are asking for a minimum of $650, double the recommended price.

rumorsvideo cardsNVIDIAGeForce RTX 2060

Comment ​Related news

Kit Guru

September 1, 2021

NVIDIA GeForce RTX 3060 and RTX 3060 Ti graphics cards will be back in limited supply in September. This is reported by the Chinese store IT Home.

This message is sure to disappoint anyone who dreamed of buying a new video card, seeing how their prices slowly stabilized, and shipments in the second quarter rose to 123 million units.

NVIDIA RTX 30 9 family0004 The site notes that shipments of the RTX 3060 and RTX 3060 Ti will drop by 50% in September compared to the first 20 days of August. This information is also confirmed by representatives of many manufacturers who are actively discussing the current situation on the forums. The situation with limited supplies will last at least until the end of September, after which, gradually, it should begin to improve.

The VideoCardz site believes that the problem will not be limited to the above two models, but will spread to the entire production of cards from NVIDIA and AMD. The reason for all this shortage is called another reduction in the production of video cards in China due to new lockdowns caused by COVID-19.

rumorsAMDNVIDIAGeForce RTX 3060 Ti3060

video card market According to recent information, video cards based on them will consume 400-500 watts of energy and at the same time provide a performance increase compared to the current generation by a factor of two.

Ada Lovelace GPUs are said to provide the same performance boost we saw from Maxwell to Pascal, and it’s been fantastic. Everyone remembers how good the GeForce GTX 9 was80 Ti, but the GeForce GTX 1080 Ti was a big technological breakthrough.

NVIDIA Ada Lovelace

Between now and the release of the 400-series graphics cards, we will have an update to the GeForce RTX 30 Ampere-series, which will receive the “SUPER” postfix. This should happen before the end of this year or at the very beginning of the next. As for the GeForce RTX 40 graphics cards with the Ada Lovelace processor, it should be expected in the second half of the year.

Rumors Lovelace Video Cards NVIDIA GPUs

Comment ​Related news

TweakTown

July 28, 2021

Next generation NVIDIA GPUs, Ada Lovelace, have been on everyone’s lips in recent days. It seems that the company has finished the design phase of the GPU, which means it’s time for rumors about its characteristics.

So, the AD102 processor will become the flagship of the next consumer generation of NVIDIA video cards. It will be manufactured in 5nm at TSMC’s factory. The chip will contain 18,432 CUDA cores. For comparison, GA102 in RTX 3090 has 10,496 CUDA cores. The expected clock speed of the new GPU will be 2.2 GHz or higher, which will provide computing performance at the level of 81 teraflops. Again, for comparison, the RTX 3090 is 35.5 teraflops.

NVIDIA

GPU The biggest change will be the use of Micron’s new GDDR6X 24Gb/s video memory, which is noticeably faster than the current 19.5Gb/s memory in the RTX 3090. Since the AD102 has a 384-bit wide bus, the total throughput will be 1152 GB/s, which is 23% more than NVIDIA’s current flagship.

The next stage in the preparation of the Ada Lovelace processor will be its pilot production. It is unlikely that a test sample will be ready before the end of the year. The final product should be expected in the fourth quarter of 2022. Given all the changes, we can assume that the performance of the future GeForce RTX 4090, compared to the GeForce RTX 3090, will double.

rumorLovelacevideo cardsNVIDIAGeForce RTX 3090

0616

When it comes to the notebook market, consumers always want the best technology at the right price. Therefore, NVIDIA GTX 1050 Ti and GTX 1060 video cards remain the most popular. But soon everything can change thanks to the GeForce RTX 3050 Ti mobile accelerator.

This video card will certainly appeal to anyone who wants to buy a gaming laptop for relatively little money. And now Videocardz has published a screenshot of GPU-Z with information about this card. As you can see, the graphics card has a peak frequency of 1485 MHz and contains 4 GB of GDDR6 video memory. This information is hardly extensive, but it’s safe to say that while this will be one of the simplest NVIDIA mobile cards on the market, it will provide sufficient performance in most games.

It’s not yet known when this graphics card will be available, but reviewers expect it to be available for purchase in the summer.

GPU-Z specs

rumorsvideo cardsnotebooksNVIDIAGeForce RTX 3050 Ti

comment on similar news Moreover, specifications and even benchmarks of the GeForce RTX 3080 Ti video card, which is based on this processor, have appeared.

The GA102-225 GPU is expected to contain 10240 CUDA cores. On the board, it will run with 12GB of ultra-fast GDDR6X memory. And such a configuration will provide unthinkable ETH mining performance — 119 MH / s. This is even more than the GeForce RTX 3090 Founders Edition and MSI GeForce RTX 3090 SUPRIM X provide, which have mining performance in the 95-115 MX/s range.

NVIDIA GA102-225 GPU

Tested RTX 3080 Ti ran at 1365MHz base GPU and 1665MHz Boost, lower than GeForce RTX 3080 and GeForce RTX 30

NVIDIA is rumored to be working on new GPUs codenamed GA106-302 to replace the GA106-300 found in the RTX 3060. These new chips will also have a new PCI Device ID, which means they won’t be able to work with older drivers and the mining cap should work again.

GeForce RTX 3060

In addition to fixing the problem of the leaked unlocked driver, the chip will also get “further mechanisms” to limit Ethereum mining on the RTX 3060. The same solutions will be used in the future on the RTX 3080 Ti and RTX 3070 Ti graphics cards .

NVIDIA hasn’t announced anything yet. If the above information is correct, then the GA106-302 GPU will be in the news in a month.

rumorminingvideo cardsNVIDIAGeForce RTX 3060

comment on similar news

KitGuru

The 100th builds everyone!

12 years ago

When choosing competitors for the GTX 580, the main thing is to stop in time. Well, the GTX 480 is exactly right for comparison — how else can you evaluate success if you don’t put the previous very best video card next to it? And then there are some questions. Compare GTX 480 with HD 5870 or not? It’s a lie, don’t go to a fortune teller, but still, at the moment it’s the single-chip flagship of the Reds (yes, I’m from old memory), you need to know how the strongest from one camp feels in the ring with the strongest from another. For balance, then you need 59Harness 70 — well, it will help to decide whether to take the second 5870, and in general, if the 5870 is a uniprocessor flagship, then the 5970 is a multiprocessor flagship. Let the «two-headed», but the most productive creation of AMD, sold «in one piece» . .. And so on.

Therefore, for the sake of simplicity, I took three more cards: Radeon HD 5870, GeForce GTX 480 and GeForce GTX 470. You can see their frequencies, video memory size and other characteristics in the table.
Yes, that’s right, it was easier to refuse overclocking this time. Firstly, this would require a change in cooling, and secondly, the GTX 580, as mentioned in the first part, has tricky voltage control, so it didn’t work out to “add” volts to the core for more speed on the go, in half an hour. Thirdly, if you overclock, then at least GTX 480 and HD 5870, and this is time again. So they put it off for the future, deciding that if you do it, then do it properly, or it’s better not to do it at all.

Eight most popular applications were selected from the variety of test applications. Of the «relevant» ones, perhaps, only the new shooter «Metro 2033» remained out of work due to extremely inadequate results for NVIDIA video cards. Given the normal performance in other applications, we can assume that the problem is either in the game or in the benchmark. Graphics settings in all games were set to maximum, eight times anisotropic filtering and four times multisampling were activated. There are only two resolutions — 1680 x 1050 and 1920 x 1080, the most common among the people. I did not set 1280 x 1024 because of the class of video cards being tested — it is unlikely that the owner of a high-end video camera will look at an «office» seed with such a low resolution. And 2560 x 1600 was not included in the tests due to its low prevalence — monitors capable of displaying a picture with this aspect ratio are still prohibitively expensive. And those who can buy them are unlikely to read this article. Well, having finished with the formalities, I turn to the description of the results.

Battlefield: Bad Company 2
Radeon HD 5870 in this test barely catches up with the GTX 470, there is no need to talk about the former and current single-chip «green» leaders. It is noteworthy that with increasing resolution, the performance drop for the GTX 480 and especially the GTX 580 is much greater than for the HD 5870. And it’s not at all a fact that a single GTX 580 will be able to provide an acceptable level of performance at a resolution of 2560 x 1600 — someone will get 40 average fps not enough.

S.T.A.L.K.E.R.: Call of Pripyat
The results of all video cards in this game could be called ideally indicative for NVIDIA: no sharp dips or jumps, the difference between the speed of cards in two resolutions is approximately the same, and the participants are lined up in the right order from a marketing point of view : at the very bottom of the HD 5870, it barely overtakes the GTX 470, and the first and second places are occupied by the GTX 580 and GTX 480, respectively. True, the owners of the GTX 470 will most likely have to give up some beauty: 40 average fps is clearly not enough, especially given the abundance of transitions from dark corridors to vast open spaces. But I do not advise turning off anti-aliasing — «ladders» terribly spoil the overall impression of the game.

Crysis Warhead
«Heavy» Crysis is so heavy… Only GTX 580 owners and monitors with a resolution of 1680 x 1050 can guarantee a good impression. If you stop worrying about every jerk, then the 480 will roll, but nothing more. The HD 5870 and GTX 470 provide a conditionally playable mode without lowering the average fps bar below 30. The owners of the last two video cards will definitely have to abandon the Ultra High mode in the graphics settings or seriously overclock — with a change in cooling, increasing the voltage on the GPU, and so on by the list.

Tom Clancy’s H.A.W.X. 2
NVIDIA strongly recommended that all reviewers include the newly minted air arcade (yes, «-arcade») in the list of test applications. The reason for her exhortation is clear from the first glance at the graph: the GTX 580 is gaining 2.5 times more fps than the HD 5870, and this is regardless of the resolution! It would be correct to say that it is reasonable to use this benchmark only to compare the performance of NVIDIA video cards of the 400th and 500th series, and Radeons, even the newest ones that are just coming out in mid-December, have nothing to do here. At all.

Call of Duty: Modern Warfare 2
We can say that MW2 is a very tolerant game: no matter what video card you stick into the slot, the fps indicator is everywhere, if not exorbitant, then “comfortable ++” (that is, with a good margin). I guess it’s the light engine combined with good gameplay that made this toy so popular. If we disregard comfort and evaluate only numbers, then the GTX 470 must be recognized as an outsider, the Radeon HD 5870 is second from the end (finally!), Well, 480 and 580 share the second and first places, respectively.

Lost Planet 2
The first version of this game was often used by Californians to demonstrate what kind of bullshit this Radeon HD 2900 XT is — the then leader, 8800 GTX, was ahead of it at times. Then, of course, the situation improved, but the brainchild of NVIDIA, in spite of everything, overtook the Reds in this game. The second part inherited the tastes of its predecessor (which is not surprising, the engine has not changed. — Approx. K.O.). And the innovations… Tessellation, of course! Now it is fashionable to have tessellation in the game. I saw one funny video about LP2, where the presenter of some presentation jumped at the screen and poked it with a pointer: they say, here is tessellation, here is a little tessellation, and here, look, look, solid tessellation! Of course, he didn’t say it directly, but … You understand (smile). Then I watched this video in better quality and realized that even carefully sticking into a 720p image (I earnestly ask the editor not to cut out the last three words — the expression is clumsy, but sensible) without indicating where to look, I don’t see megatessellation at all …

A little distracted. So, what do we have with the results… Well, everything is clear: the owners of HD xxxx are assigned resolutions from 1680 x 1050 and lower, without anti-aliasing. Even if we assume that a glitch has crept in somewhere in the results, and add another ten fps to 5870, the picture will not change much. Having fun shooting alien creatures at maximum quality settings is possible only for owners of the eighties, those from NVIDIA. The rest will have to lose in quality and / or resolution.

Far Cry 2
To be honest, I never understood what fans find in this game. Video cards, on the other hand, reacted with humor, especially 470: the difference of 35% between the indicators in two resolutions against the background of 10-15% for the other three cards (two of which are older brothers 470 in terms of architecture) looks strange. However, this does not prevent the former and current flagship NVIDIA from sharing the second and first places, placing 470 in third place, and giving an outsider’s third place to the opponent from the «red» camp. If we forget about the numbers and look at the picture as a whole, we can see that it doesn’t really matter which of the four tested video cards is working on the game. The minimum 75 fps shown by the Radeon HD 5870 is enough for a comfortable game.

3DMark Vantage
Futuremark’s test/synthetic/game suite was the second of eight apps to rank the HD 5870 third instead of fourth overall. But, as has been repeatedly noted, it is not worth transferring the balance of power in parrots to those in efpees, there is a great chance of making a mistake. Old-timers can remember the events of five or seven years ago, when the number of approaches to building a picture was much smaller and the largest number of parrots guaranteed success in all more or less modern games with almost one hundred percent probability. Why go far for an example! Look at the gap between the GTX 480 and the GTX 580. And now for all the other charts… How many games show that difference (percentage)? With an interference — four. It can be said that the two remaining ones, Far Cry 2 and MW2, are too old in their essence to adequately “appreciate” the innovations in the 580, but then what about H.A.W.X. 2 echoes them? A BF2 at 1920 x 1080? Here. So…

Conclusions
…Advice: if you choose a high-end solution not for show off, but for good playability, make a list of the games you use most often and evaluate performance in them. Unfortunately, it will not be possible to turn this with toys that have not yet been released, be guided by your own wallet. Now for each video adapter in detail. The GeForce GTX 580 lived up to the expectations placed on the new architecture. The only downside is that it came out much later than planned, but there is also a plus — the GTX 480 made it possible to identify flaws and, if possible, fix them by the release of the 580. At least that’s how it looks from the point of view of an ordinary user. You can philosophize and argue a lot about the behind-the-scenes struggle, the efforts of marketers, and the repeated skimming of the same technology, but when you go to the store, these arguments, alas, will not bring any benefit.

Should I buy 580 now? See why. It is definitely not worth changing 480 for it — the end will not justify the means. And those who are not interested in rustling paper banknotes are unlikely to think about buying for a long time. The advantage of the GTX 580 over AMD’s current single-chip flagship is significant, mainly due to architectural changes and improvements. If you are a fan of AMD products, you should wait for the announcement of 6970 and 6950 — who knows, suddenly the new product will turn out to be no worse, or maybe even better … the number of new products does not contribute to the installation of price tags corresponding to the actual ranking of video cards in terms of strength.

The

GeForce GTX 480 is still strong and shows adequate results — there are very few games that «poorly ran» at 480, but at 580 they do well. In my personal opinion, it’s easier to buy a second 480 right after the New Year and assemble an SLI pair. She will eat a lot, bask too, but in return you will get a tandem on which you can feel comfortable until the release of the successor 580 (read — another year). By the way, the water cooling system helps a lot in the fight against the hot temper of the 480. The one from the category of high-performance. It will cost the owner 500-1000 bucks, but, as practice shows, with the right choice of components, it will survive two or three upgrades without any problems.

The

GeForce GTX 470 is a decent gaming graphics card: moderately hot, moderately productive. The optimal combination of «price-performance» indicators for the middle segment. Someone may object that, they say, 460 bypasses 470 in this parameter. I will answer: this is a matter of personal preferences and desires. The Radeon HD 5870 is already rather weak, its replacement arrived just in time. In principle, you can buy another 5870 and assemble CrossFire X — this will help to hold out for another year. The performance level will be roughly comparable to the agility of a pair of GTX 470s and will greatly lose to a bunch of two 480s, but both the appetites and the price of such a solution will be much lower. Here is such a schedule. It can be shifted in one direction or another at will, changing the frequencies of the cards in both directions, upgrading the cooling, assembling various tandems, buying low-used cards (this is sometimes a very profitable move from a financial point of view). In general, you have points of reference, then be guided by personal preferences and the amount of funds available. U.P.

How to measure gaming performance?
A rare game maker cares about the needs of card testers. What did not go to pay for designers, artists and programmers is pumped into the marketing campaign: it is much easier (and cheaper!) to convince the user that the toy is good than to allow him to come to this conclusion himself (after making the game good, of course) . Therefore, built-in tools for measuring performance are not everywhere. In the list of «well done», for example, S.T.A.L.K.E.R. — it allows you to record and then play the demo launched from the console. For some games, such as Unreal Tournament 3, Crysis, Far Cry 2, third-party craftsmen write benchmark programs that carry out the entire routine process for the tester.

And for other games that do not have built-in or external means for measuring «nimbleness», the Fraps utility is popular (it calculates the average fps) and scripts that repeat the route after the player. And then, scripts are a boon available only to those who understand programming (here I wanted to write to “users”, but stopped in time) and to those who are able to pay them money for writing a script. The rest from time to time sit in front of the PC, repeating the movements of the mouse. Badly? Yes. But even worse, the error in the results produced by Fraps can vary by 10-15%, and sometimes by 20%, from run to run. Well, if the script runs everything: I set it to run the game ten times, went for a walk, came in, collected the results and, say, out of ten passes, threw out two or three with abnormally high or low numbers. And if you walk with your “hands” every pass, then you can’t master it more than two or three times, time is precious.

But, as they say, progress does not stand still, and, despite the dominance of game consoles and excellent projects for them, many continue to torment the PC with games. Consequently, there is interest in graphics cards and reviews like the one you are reading now, and there is no end to trying to make a thing that makes it easier to measure performance in games.
P. S. The last paragraph was written by the author after a sleepless night and a battle with Metro 2033 in the hope that someone will hear him and give a link to the bench of this demonic fun.

Review of MSI GeForce GTX 580. The top video card of today

Introduction
In today’s review, you will be presented with the top video card of our time — NVIDIA GeForce GTX 580. This video card can justifiably be considered a solution for enthusiasts, since such a high level of performance unlikely to be needed by the average gamer, but absolutely essential for the gamer who wants to play at resolutions above 1920×1280. Of course, if the user has the opportunity, then why not buy a top-end video card and forget about the need to replace it when a new gaming product appears?

Actually, there is a reason for this. As a rule, the cost of top solutions is at the highest level. In six months, new video cards will be released that will replace the former top product from the top price line and its cost will be in the middle price range. Approximately the same fate befell the top-end video cards of the past generation — NVIDIA GeForce GTX 480, which today demonstrate performance at the level of a video card from the middle price range — NVIDIA GeForce GTX 570. Now, those who want to purchase the former top-end product from NVIDIA have an alternative with the opportunity to spend on product up to 40% less funds.
To date, AMD has only one solution related to the previous generation of graphics products, which it can oppose to the top-end solution from NVIDIA presented at the end of last year. This product is called AMD Radeon HD 5970. This graphic solution is based on two graphics cores, and for quite a long time it was the most productive video card on the computer component market. Due to the need to separate graphics calculations between two video cards, the solution periodically had certain problems called friezes — periodic twitching of the image during the game. AMD’s detailed work with graphics card drivers has eliminated this problem and the demand for the previous generation of AMD Radeon HD 59 graphics cards70 still remains at a fairly high level.

The absence of a top product haunted the engineers of NVIDIA for several months and in November 2010 they present a new video card — NVIDIA GeForce GTX 580. This is the first graphics product that fully uses 40 nanometer process technology and Fermi architecture. The graphics card has an increased number of stream processors from 480 to 512 compared to the graphics solution of the previous generation NVIDIA GeForce GTX 480.

From a marketing point of view, the launch of the new GeForce GTX 580 was not in vain in November 2010. With its new two products GeForce GTX 580 and GeForce GTX 570, NVIDIA has seriously competed with existing solutions from AMD and flooded the world markets with new products by the beginning of New Year’s purchases. Who doesn’t want to get a new graphics card for the new year? — only to those who do not like to play computer games, and there are practically none today. It is quite possible that the product was prepared for release quite a long time ago, and the effect of surprise and the sales season were not enough for the announcement.

AMD only in January 2011 introduced its new AMD Radeon HD 6970 and Radeon HD 6950 , which could not directly compete with the top-end solution from NVIDIA. These solutions adequately compete with weaker products — the GeForce GTX 570 and GeForce GTX 560 Ti, and only the dual-chip AMD Radeon HD 5970 remains to compete directly with the TOP.

In today’s review, we will try to fully evaluate the capabilities of the new product from NVIDIA in the face of the MSI GeForce GTX 580 graphics solution.

Equipment

— The picture is clickable —

The video card is delivered in a blue and white box. On the front side of the box, the software supplied with the video card is advertised — MSI AfterBurner, which allows you to control the parameters of the video card and overclock it with the accompanying softvolt mod. Enthusiasts have the ability to manually adjust the fan speed of the cooling system — we have never noted an urgent need for this function.

The top bay lists the technologies supported by the graphics card. This set is not much different from the technologies of the previous generation video card — GeForce GTX 480. It should only be noted that there is 1536 MB of GDDR5 video memory on board. This circumstance requires criticism from our side. New video cards from AMD carry 2 GB of video memory on board, which can have a very sharp impact on the level of sales of video cards. Many modern users still choose a video card based on two characteristics:

— cost,

— the amount of video memory.

The lower the cost of a video card and the higher the amount of its video memory, the more chances this solution has to be sold.

— The image is clickable —

On the back of the video card, the manufacturer lists the key features of the solution. At the same time, MSI focuses on its innovations in this product. In particular, the presence of MSI AfterBurner software, which works with almost any video card from the NVIDIA and AMD series. Working with the program is quite easy and convenient, especially for those who already have experience with the free Riva Tuner utility.

The company notes the features of the capacitors used in this graphic solution. They are all solid-state with an increased resource, which serve up to 40 years in an office environment. In fact, an office computer does not need such a video card in any way, unless there is a web studio or a design studio there, even in such extreme cases, you can get by with more modest graphic products or solutions from the professional segment — Quadro.

— Clickable image —

Delivery includes:

— video card,

— instruction,

— two power adapters,

— driver disk,

— DVI-to-VGA adapter,

— mini-HDMI-to-HDMI cable.

The inclusion of an HDMI cable in the package is a completely justified step, although its length is unlikely to be enough to solve any significant tasks, except for connecting a monitor. But he is able to demonstrate the presence in nature of cables with mini-HDMI ports at one end and HDMI at the other end. Visual inspection of the video card

— The picture is clickable —

MSI branded graphics card has a full reference design. The reference PCB design is confirmed by the NVIDIA inscription in the area of ​​the PCI-Exp slot. The cooling system cover is marked from the delivery box, however, it does not differ in any way from the original NVIDIA solution.

If we compare this cooling system with the one installed on the previous generation video card — GeForce GTX 480, then in this solution you will not see heat pipes. The new cooling system retains only the cooling turbine, which blows air through the radiator and expels it outside the case. The radiator of the cooling system has undergone significant changes. First of all, the changes relate to the exclusion of heat pipes and the construction of the entire architecture on the principle of Vapor-X cooling systems from Sapphire. That is, between the copper base of the cooler and the radiator itself there is a heat distribution chamber, which performs the function of heat pipes. Only this function is performed more efficiently and at a higher level than the use of standard heat pipes.

The turbine of the cooling system is equipped with a PWM speed controller. At low loads on the video card, the turbine speed is insignificant and its operation is not audible, with a gradual increase in load, the turbine speed increases with a simultaneous increase in operating noise from air flows. In general, the original cooling system from NVIDIA cannot be called noisy.

— Clickable image —

The video card has two connectors for SLI bridges, which allows you to combine up to three video cards into a single tandem. Naturally, for this, your motherboard must support the installation of so many video cards and 3-way-SLI technology.

— Clickable image —

At peak loads, the power consumption of the video card reaches 244 watts, so to ensure compatibility with PCI-Exp 16x slots of the first revisions, two connectors are installed for additional power supply of the video card. One connector is six-pin, the other connector is eight-pin.

On these video cards, a standard set of ports for image output is soldered: 2xDVI and 1xmini-HDMI.

Video card specifications

1. Working frequency of universal computing units: 1544 MHz;

2. Universal computing units: 512;

3. Texture blocks: 64;

4. Blending units: 48;

5. Core operating frequency: 772MHz;

6. Video memory size 1536 MB;

7. Video memory type: GDDR5;

8. Video memory operating frequency: 1002 MHz;

9. Memory bus width: 384-bit;

10. Maximum power consumption up to 244 watts.

The MSI product presented to your attention operates at reference frequencies, although some time later the company announced solutions with a modified cooling system and factory overclocking. We would like to hope that we will be able to present these products to your attention in the next reviews. Test configuration

The MSI GeForce GTX 580 video card was installed in our working configuration, which, although not top-notch, is quite capable of claiming the title of a modern configuration:

1. Intel Core i7 920 processor.

2. ASUS P6T motherboard.

3.2×3 Gb Samsung Original DDR3-1600

4.WD 1TB WD1001FALS Caviar Black SATAII

5. Thermaltake Mambo case.

6. ASRock USB 3.0 PCI-Exp x1 dual port board.

7. The room is 27 degrees.

8. The system is assembled in a closed case.

Based on this configuration, all graphic solutions presented below in the review were tested. An OCZ power supply for 9 was used as a power source.20 watts.

1. The temperature regime of the video card.

From the presented test results, it can be seen that the reference cooling system for the GeForce GTX 580 video card, although it does not claim to be the most efficient, can quite cope with the cooling of a video card operating at reference frequencies.

It should be noted a sharp increase in cooling efficiency compared to the cooling system of video cards of the GeForce GTX 480 series, which even then worked on the verge of overheating after a slight overclocking or when another hot summer in our country sets in.

During the softvolt mod, the voltage on the core was increased to a maximum of 1.15 volts, which led to an increase in the overclocking potential of the video card with a simultaneous increase in the maximum temperature to 97 degrees. An increase in temperature to 92 degrees led to a sharp decrease in the overclocking potential of the video card, so the most effective voltage for the video card turned out to be 1.13 volts.

It should be understood that any volt mod must be accompanied by adequate cooling. Previously, it was proved that the lower the temperature of the core, the better it responds to an increase in voltage by increasing the overclocking potential.

2. Video card overclocking

Any software can be used to overclock the MSI GeForce GTX 580 graphics card. The most convenient program seems to us from the package — MSI Afterburner.

By default, the graphics card operates at the following operating voltages:

— 0.96 volts in 2D mode,

— 1.05 volts in 3D mode.

Without volt mode, the video card overclocked to the following frequencies:

— 850 MHz for the core,

— 1174 MHz for video memory.

Increasing the core voltage to 1.13 volts made it possible to achieve the following frequencies:

— 894 MHz for the core,

— 1174 MHz for video memory.

3. Evaluation of the performance level of the video card in the game Crysis Warhead.

A fairly popular game, which is an objective criterion for the level of performance of a video card today.

The test results show that the tested video card GeForce GTX 580 can be considered top only among single-core solutions. The performance leader in this gaming test remains the dual-core solution from AMD Radeon HD 5970.

If we compare the product with the previously released solution GeForce GTX 480, it should be noted that the new product has become more productive, while it began to consume less energy and work more quietly.

4. Video card testing in Resident Evil 5

This game test clearly demonstrates that the effectiveness of a dual-chip solution can only be evaluated in a game that supports such video cards. In this case, we see the absence of this support, so the graphics product from NVIDIA — GeForce GTX 580 is the leader in testing.

5. Video card testing in Far Cry 2.

This game is a popular product and it fully supports many modern video cards. Traditionally, it prefers graphics solutions from NVIDIA, which is clearly seen from the presented performance charts. Top solutions from AMD and NVIDIA demonstrate approximate parity in terms of performance.