Cas on ram: Evaluating Standalone DRAM Performance — Memory Frequency Scaling on Intel’s Skull Canyon NUC

Evaluating Standalone DRAM Performance — Memory Frequency Scaling on Intel’s Skull Canyon NUC

by Ganesh T Son August 29, 2016 8:00 AM EST

  • Posted in
  • Systems
  • G.Skill
  • Intel
  • Corsair
  • Crucial
  • Kingston
  • Patriot
  • DDR4
  • NUC
  • DRAM
  • Skylake-H



IntroductionPremium DRAM Options for the Skull Canyon NUCEvaluating Standalone DRAM PerformanceCPU Performance BenchmarksGPU and Gaming BenchmarksMiscellaneous Aspects and Concluding Remarks

Intel’s Memory Latency Checker tool can measure memory latencies and bandwidth, and how they change with increasing load on the system. It also provides several options for more fine-grained investigation where bandwidth and latencies from a specific set of cores to caches or memory can be measured as well. The tool also disables the sophisticated hardware prefetchers in order to give an idea of the actual performance of the tested component (cache or main memory).

Most of the test options available in Intel’s Memory Latency Checker tool are an overkill for / not applicable to systems like the Skull Canyon NUC. In this section, we present selected main memory benchmarks for the various tested kits.

Idle DRAM to DRAM Bandwidth

Idle DRAM to DRAM Latency

Peak DRAM Bandwidth — 1:1 Reads:Writes

Peak DRAM Bandwidth — 2:1 Reads:Writes

Peak DRAM Bandwidth — 3:1 Reads:Writes

Peak DRAM Bandwidth — All Reads

There are no surprises at all in the bandwidth graphs — faster the memory frequency, the higher the bandwidth, whatever the scenario. Latency is a slightly different story, with certain high frequency kits with larger (worse) timing parameters performing better than kits with better timing parameters, but operating at a lower frequency. It is clear from the above graphs that if one has a purely main-memory-bound (not cache-sensitive) workload, it would benefit immensely from the G.Skill Ripjaws 3000 MHz kit. While the bandwidth numbers are excellent and according to expectations, even the actual latency numbers (in terms of ns rather than clock cycles) are better than the Kingston HyperX kit. Are there any real-world applications that can benefit from this performance? The next couple of sections provides the answers.

Premium DRAM Options for the Skull Canyon NUC
CPU Performance Benchmarks
IntroductionPremium DRAM Options for the Skull Canyon NUCEvaluating Standalone DRAM PerformanceCPU Performance BenchmarksGPU and Gaming BenchmarksMiscellaneous Aspects and Concluding Remarks


What Are Memory Timings? CAS Latency, tRCD, tRP, & tRAS (Pt 1) | GamersNexus


What Are Memory Timings?

First, the basics. While memory frequency is measured in Hertz, or cycles per second, the unit for memory timings is just plain cycles. To convert clock cycles to a measurement of time requires knowing the frequency of the memory. This is listed in MHz, or units of 1,000,000Hz. 3200MHz memory has a clock frequency of 3,200,000,000 cycles per second, so the time for a cycle to complete should be (1/3,200,000,000) seconds. However, modern memory is DDR (double data rate), meaning data is transferred on the rising and falling edge of each clock, so advertised frequencies are twice the real clock frequency. That’s why when you set memory to 3200MHz in BIOS, CPU-Z will show 1600MHz. Therefore the time is really (1/(3,200,000,000/2)) seconds. If our example memory has a CL timing of 16 clock cycles, this translates to (16 * (1/(3,200,000,000/2))) seconds, or .00000001 seconds, or 10 nanoseconds.

The equation is (1/(advertised frequency/2)) * timing in cycles = timing in seconds. For a DDR3-1600 kit, a CL of 9 translates to 11.25ns, actually slower than our previous example. Latencies have gradually increased over the years with the physical distance that signals have to travel (the speed of light is a hard limit), but frequency has increased as well, and therefore performance has still improved. Frequency is very important, but it’s just one element of performance, as with CPUs.

There are many, many different timings, but they deal with a fairly small list of commands: when they can be issued, how long it takes for them to execute, how many cycles pass before a response. Here’s a table of DDR4 commands copied from Wikipedia:

The important signals from the top row are ACT (Activate), RAS (Row Access Strobe), CAS (Column Access Strobe), and WE (Write Enable). RAS and CAS are often referred to simply as column and row address because they aren’t actually strobes; the terminology is a holdover. These are Active-Low signals, so they can be either H(igh) or L(ow), 1 or 0. Together they form a four-bit code that specifies a command to be executed. The signals have changed a bit over the years, but for the most part DDR4 has the same list of commands that SDRAM has always had, and therefore many of the same timings. For some background information on stuff like banks, rows, and columns, this venerable 2010 article from Anandtech is a good overview of what SDRAM actually is and how it functions.

For this next section, we created a custom animation that can be found in our above-embedded video. That may assist in better understanding the below definitions.

Activate: opens a row of a bank. A row must be active for reading and writing data. If a row isn’t active it’s idle, and if a row is activated it stays that way until a precharge command.

Precharge: closes the open row in one or all banks (two separate commands), putting them into the idle state. Data is still stored in idle banks, but they must be activated again before reading or writing.

Read and Write: self explanatory. With these commands, an Auto Precharge flag can be set to automatically precharge the row when done.

ZQ calibration: compensates for temperature/voltage variation. It can be a recurring command, but not frequently enough to make the related timings important to us.

Refresh: refreshes the charge in memory cells by writing data back in place without changing it. DRAM is volatile memory, which means that it requires power to store data: bits are represented by charges on capacitors, which leak over time if they aren’t read or written to. We’ll discuss this more in the next article: look for tREFI and tRFC. All banks must be idle (precharged) before a refresh.

Timings are generally divided into three categories: Primary, Secondary, and Tertiary. Primary is the broadest, the rated settings are listed on the box (e.g. 16-16-16-32 2T), and they’re usually adjustable in BIOS. Secondary are non-primary timings that can optionally be set in SPD (see next section). These include: tWR, tRFC, tRDD_L, tRDD_S, tWTR_L, tWTR_S, tRTP, tFAW, and tCWL. Their names and definitions will be covered in a future article. Tertiary is the catch-all category for everything else, and it’s a deep rabbit hole—whether or not these settings are even exposed in the BIOS varies heavily with motherboard quality, and the values vary based on individual RAM, CPU, and motherboard. We’ll be covering these further in another article.

SPD and XMP Explained

That last paragraph requires some additional explanation. Every stick of RAM contains information baked into EEPROM (Electrically Erasable Programmable Read Only Memory), including some safe timing/frequency profiles—these can be viewed in the SPD tab of CPU-Z under the JEDEC header.

The JEDEC (Joint Electron Device Engineering Council) Solid State Technology Association is an organization that publishes standards for DDR4, DDR5, SSDs, mobile memory, ESD, GDDR6, and more. They are responsible for standardizing and defining everything in this article, from abbreviations to the entire concept of DDR4. As part of this, JEDEC publishes tables of baseline timings for different RAM types, several of which are saved in SPD. Memory manufacturers technically produce strictly JEDEC-defined varieties of RAM, like DDR3 1600MHz 11-11-11-28, but can make them capable of higher speeds and market them as such. For example, Corsair sells 4600MHz memory, which as of this writing is a higher speed than JEDEC has defined for DDR4. When a newly-built system is powered on for the first time, the board will check SPD and default to the best set of these JEDEC approved slow-but-safe speeds.

XMP is an SPD extension. XMP, or eXtreme Memory Profiles, are higher-performance specs jammed into the leftover space in the EEPROM. These are conceptually the same thing as the JEDEC profiles, but they’re optimized by the memory manufacturer. XMP contains settings that the memory manufacturer says will probably work, but may not be supported depending on the quality of the CPU IMC or SOC and other components. For enthusiast RAM, the advertised speeds are usually only attainable by applying XMP.

XMP is technically an Intel term and an Intel standard, but an “extreme memory profile” is literally just a list of numbers that can be read by any system (if the board allows it). For example, DOCP and EOCP on AMD boards are just generic names for XMP. “XMP Profile” is redundant, but just remember: nobody cares, including Intel, or they’d have abbreviated “extreme” correctly.

Neither SPD nor XMP contains every timing. According to a Kingston representative: “We tune the ‘Primary’ timings (CL,RCD,RP,RAS) only. Other timings are left alone at JEDEC’s recommendations, likely for the MRC.” In regards to some specific subtimings we asked about, “since they are not in the SPD (or XMP), we can’t change them even if we need to.” This lines-up with our experiences in the “Real World Impact” section of this article. Even if memory manufacturers wanted to go deeper, there is a specific and limited list of SPD entries.

If you’re reading these articles because you want to know whether to tune memory, the answer for most mainstream users is “no, definitely not.” We’ve done a lot of memory tuning for our #RIPLTT streams, and it can take a day (or more) to dial-in optimal timings — and even those might have occasional memory errors. Enabling XMP and calling it a day generally does a good enough job. Memory tuning is the domain of overclockers, hardware enthusiasts, and people with a lot of time on their hands, but it’s not worth it for users who just want something with low effort requirements. XMP will cover that camp.

What Is Memory Training & How to Train Memory?

So, when it comes down to it, RAM doesn’t set its own frequency and timings—the BIOS does, with SPD/XMP as a starting point. On Intel boards, this element of the BIOS is called the MRC. This is one area where board manufacturers can secure an advantage in performance, because the sheer variety of RAM and the differences between Samsung, Micron, and Hynix chips make tertiary timings hideously complex to adjust. Memory and board manufacturers can work together to bake-in optimal timings for popular kits, but for the most part these are determined (if left on auto) during POST, where they should remain unchanged unless there are boot failures.

Memory training can appear as a black art to casual overclockers, where RAM can be magically “broken in” by repeatedly forcing boot failures until an unstable OC becomes stable enough for benchmarks. What’s actually happening is that the IMC will try a variety of settings in an attempt to stabilize the system (not all of them timings). On the user end, it can feel like trying to solve a jigsaw puzzle by shaking the box, but it works. This is an important point even for non-overclockers: if memory isn’t stable with XMP, allow the system to reboot a few times before giving up.

Real-World Impact of Lower Level Memory Timings

Different kits have different JEDEC standards, so different subtimings get loaded. This is an issue we’ve run into before: we had two kits identical in frequency and capacity, but with different latencies; however, the 16-18-18-36 Corsair kit somehow outperformed the 14-14-14-34 G.SKILL kit in tests. After discussing this with both manufacturers, it turned out that Corsair’s sticks were dual-sided with groups of 512MBx8 ICs (memory chips), while G.SKILL’s were single-sided with groups of 1GBx8 ICs.

512MBx8 is an older style, so it actually has tighter subtimings defined by JEDEC (see “What Are Timings?”). The biggest culprit in this case was tRFC, or Refresh Cycle Time. We’ll cover the full definition of this in part 2, but for now the important thing is that the value set by JEDEC for Corsair’s type of kit was 416 and 560 for G.SKILL’s. When we adjusted the G.SKILL kit’s tRFC value down to match Corsair’s with no other adjustments, it pulled ahead in performance and remained stable, as shown in the Ashes of the Singularity results below:

This is an extreme case, but it shows both the value of experimenting with subtimings and the frustrations of benchmarking memory. If the ASUS board had happened to have a set of optimized timings for this specific G.SKILL kit, it would have performed better than the Corsair kit from the start.

Primary Timings: tCL/tCAS, tRCD, tRP, & tRAS Explained

With all that out of the way, it’s time to talk about the primary set of timings. On any product listing, box, or stick of RAM, timings will be listed in the format tCL-tRCD-tRP-tRAS, with CR sometimes listed as well.

CAS Latency (tCL/tCAS):

Wikipedia: “The number of cycles between sending a column address to the memory and the beginning of the data in response. This is the number of cycles it takes to read the first bit of memory from a DRAM with the correct row already open. Unlike the other numbers, this is not a maximum, but an exact number that must be agreed on between the memory controller and the memory.”

CAS Latency is the most widely talked about and compared memory timing. The CL timing is an exact number, the base time that it takes to get a response from memory in the best possible scenario described above, referred to as a “page hit.” The other primary timings (other than command rate) are minimums. It’s important to remember that although we’ll be talking about how these timings relate to reading data from memory, that’s just one thing they affect.

RAS to CAS Delay (tRCD):

Wikipedia: “The minimum number of clock cycles required between opening a row of memory and accessing columns within it. The time to read the first bit of memory from a DRAM without an active row is tRCD + CL.”

RAS to CAS is one potential delay to read/writes. tRCD is the number of clock cycles it takes to open a row and access a column. If a request for data is made when there are no rows open, referred to as “page miss,” it will take at least tRCD + CL clock cycles for the CPU to receive the first bit of data in response.

Row Precharge Time (tRP):

Wikipedia: “The minimum number of clock cycles required between issuing the precharge command and opening the next row. The time to read the first bit of memory from a DRAM with the wrong row open is tRP + tRCD + CL.”

If the wrong row is open (“page miss”), it needs to be closed (precharged), then the next needs to be opened, then the column within the row needs to be accessed. This therefore takes tRP + tRCD + CL time.

Row Active Time (tRAS):

Wikipedia: “The minimum number of clock cycles required between a row active command and issuing the precharge command. This is the time needed to internally refresh the row, and overlaps with tRCD. In SDRAM modules, it is simply tRCD + CL. Otherwise, approximately equal to tRCD + 2×CL.”

Also known as Activate to Precharge Delay or Minimum RAS Active Time. The first equation (for SDRAM) is the relevant one here, but it should be more than that. We’ve seen multiple different “true” ways to calculate tRAS, but given the complexity of memory operations, good old trial and error remains the easiest. For example,  we somehow booted with memory at 16-16-16-26, and that doesn’t make sense by anyone’s rules.

Command Rate (CR/CMD/CPC/tCPD):

AMD: The amount of time, in cycles, between when a DRAM chip is selected and a command is executed. 2T CR can be very beneficial for stability with high memory clocks, or for 4-DIMM configurations.

Also known as Command Period. This will either be 1T or 2T on modern memory, with 1T being faster. Despite the unique -T notation, this is measured in clock cycles like the other timings. There’s generally a very small performance delta between the two options.

Conclusion of Part 1

That’s it for the basics, but there are plenty of other timings to define and explain. Check back for part 2, where we’ll explore the secondary and tertiary timings of DDR4 memory.

If any of this would be better with a visual aid, try our embedded video (above) for an animated walkthrough of how some of the memory accesses work. We’ll be back with secondary and tertiary timings soon enough.

Editorial, Research: Patrick Lathan
Host: Steve Burke
Video: Andrew Coleman

Determination of jurisdiction of cas \ Acts, samples, forms, contracts \ Consultant Plus

  • Main
  • Legal resources
  • Collections
  • Determining the jurisdiction of Cas

A selection of the most important documents upon request Determining the jurisdiction of Cas (regulations, forms, articles, expert advice and much more).

  • Administrative proceedings:
  • Administrative claim
  • Administrative claim for inactivity of the bailiff
  • Administrative claim
  • Address of the eighth court of cassation
  • administrative proceedings:
  • Administrative statement of claim
  • Administrative statement of claim for inaction of the bailiff
  • Administrative claim
  • Address of the eighth court of cassation
  • Audio recording cas
  • Show all

Judicial practice


Selection of court decisions for 2021: Article 33 «Place of execution and enforcement measures» of Federal Law No. 229-FZ of October 2, 2007 «On Enforcement Proceedings» «Based on the provisions of Article 33 of Federal Law No. 229 of October 2, 2007-FZ «On Enforcement Proceedings», paragraph 8 of the Decree of the Plenum of the Supreme Court of the Russian Federation dated September 27, 2016 N 36 «On Certain Issues of the Application by Courts of the Code of Administrative Procedure of the Russian Federation», when determining the jurisdiction of an administrative statement of claim, it is not the location of the court the bailiff, for the execution of which the executive document was received, the division of the bailiff service, in which the bailiff performs his duties, and the place of action to fulfill the requirements of the executive document, which are determined by the location of the debtor in enforcement proceedings or his property. «

Articles, comments, answers to questions

Register and get trial access to the ConsultantPlus system for free at 2 days

December 1, 1996 N 1-FKZ «On judicial system of the Russian Federation»
(Zherebtsov A. N., Kireeva E.Yu., Peshkova (Belogortseva) H.V., Baranov I.V., Mayboroda V.A., Parfiriev D.N., Strukov K.V. , Chernus N.Yu.)
(Prepared for the ConsultantPlus system, 2021) other cases may also be included.This means that the jurisdiction of administrative cases to justices of the peace is determined by the rules of the CAS, in Article 17.1 of which it is provided that justices of the peace consider applications for issuance of a writ of injunction on claims for the collection of mandatory payments and sanctions. administrative proceedings justices of the peace consider administrative cases of the given category.


Procedural coercion will be clarified

Experts supported the emergence of new clarifications on the application of the rules of the CAS RF, calling the draft resolution of the Plenum of the RF Armed Forces «balanced and convenient for practical application.»

On June 1, 2017, at a meeting of the Plenum of the Supreme Court of the Russian Federation, a discussion was held of a draft resolution on the application by courts of procedural coercive measures when considering administrative cases. Judge of the Supreme Court of the Russian Federation Alla Nazarova, speaking at a meeting of the Plenum, noted that the consolidation of the institution of procedural coercion in an independent section and Ch. 11 of the CAS RF was one of the important novelties in the legal regulation of the consideration of administrative cases.

“The Code of Administrative Proceedings provides for a completely new measure for litigation in administrative cases — the obligation to appear; establishes a procedure for imposing a judicial fine that is fundamentally different than the Code of Civil Procedure of the Russian Federation; regulates in detail the procedure and grounds for the application of such measures as: restriction in the entry of a participant into the trial, deprivation of the participant in the trial of the floor, warning about removal from the courtroom. These novelties of administrative proceedings objectively necessitated the development of a resolution of the Plenum,” the judge explained.

She added that the document was prepared based on the results of work on the study, analysis and generalization of judicial practice in the application by courts of the provisions of the CAS RF on procedural coercion.

Artem Berlin, Corporate and Arbitration Practice Lawyer at Kachkin & Partners, noted that, in general, the draft resolution is a detailed and structured retelling of the relevant provisions of the RF CAS. “It is convenient to use it in practical work, however, there are not so many novels of interpretation in it,” the expert believes.

Thus, the draft explains the concept and meaning of measures of procedural coercion, as well as the grounds for their application. The division of responsibility for non-execution of a lawful order of a judge to stop actions that violate the rules established in court, provided for by the CAS of the Russian Federation and the Code of Administrative Offenses of the Russian Federation, is explained. The definition of the circle of persons to whom measures of procedural coercion can be applied is given.

Clauses 7 to 10 contain clarifications that allow the courts to decide in what cases and in what sequence certain measures of procedural coercion are to be applied.

According to Artem Berlin, it is interesting that, according to paragraph 9 of the draft, it is allowed to remove a person from the courtroom without prior warning if the violation of order committed by him in the court session is of a significant nature. “Although this does not directly follow from Article 119 of the RF CAS, such an interpretation is logical and consistent with established practice,” he said.

The expert also pointed out that the clarification contained in paragraph 11 of the draft resolution is important enough to ensure the rights of the participants in the process. “According to it, coercive measures affecting the right to participate in the case (restriction or deprivation of speech, removal) should be appointed in accordance with the principle of reasonableness, which implies taking into account the materiality of the violation committed, the nature of the guilt of the person who committed it, and the circumstances of the violation. The court, therefore, is obliged to justify the measure it chooses and cannot apply it arbitrarily,” the lawyer explained.

Paragraphs 12 to 15 cover in detail the procedure for applying such a measure of procedural coercion as the drive and execution of a ruling on its application. “The clarifications contained in paragraph 15 of the draft concerning the resolution of issues on the recovery of expenses for a drive are of fundamental importance for practice. Unfortunately, the courts have a different practice of recovering these costs. A number of courts believe that the relevant costs are court costs in the case and, accordingly, must be recovered from the person against whom the final decision was made,” Alla Nazarova said during her speech.

She pointed out that the draft resolution reflects a different position based on a literal interpretation of the provisions of the Code. At the same time, the authors of the draft proceeded from the fact that the costs of the drive cannot be recovered from persons whose action or inaction did not entail the occurrence of such costs.

A number of paragraphs of the draft is devoted to such a new measure of procedural coercion for administrative proceedings as an obligation to appear. Artem Berlin, analyzing the document, drew attention to the fact that they allowed the application of the obligation to appear to persons who had not previously committed violations of the procedural law. “According to paragraph 16 of the draft, such an obligation can be imposed by a court ruling immediately upon acceptance of the claim for proceedings, appointment of a court session. To ensure the prompt consideration of cases, this interpretation can be very useful, ”the expert believes.

Concluding her speech at the meeting of the Plenum, the judge of the Supreme Court of the Russian Federation Alla Nazarova said that a significant number of court questions addressed to the Supreme Court of the Russian Federation were related to the interpretation of the concept of «contempt of court» and the definition of violations for which a person may be held accountable for contempt of court. The explanations contained in paragraph 19 of the resolution are devoted to the solution of such issues.

However, this paragraph aroused concern for Artem Berlin, since the concept of “contempt of court” is interpreted quite broadly in the document. “According to the draft version, it may include, among other things, ‘not due to a change in the circumstances of the case or other objective reasons, the repeated statement of the same petition, in respect of which a court ruling has already been issued and announced’. It seems that such a wording leaves room for abuse by the court, which will have the opportunity to punish participants in the process for procedural activity,” the lawyer said.

Nikolai Yashin, lawyer of the City Administration of Moscow, also generally positively assessed the development of a resolution of the Plenum of the Supreme Court of the Russian Federation, which specifies the procedure and grounds for the application of procedural coercion measures.

At the same time, he noted that the Supreme Court of the Russian Federation had previously relatively recently addressed the issue of the application of procedural coercion measures in the framework of administrative proceedings — the relevant clarifications were given in September 2016 in the framework of the Decree of the Plenum of the Supreme Court of the Russian Federation No. administrative proceedings».

“The attention currently paid to procedural coercion measures in administrative proceedings is justified for the purposes of the uniform application of these norms by the courts, which is what the document says. The possibility of imposing such measures carries a large share of the ‘discretionary’ element, which may lead to judges abusing their rights. On the other hand, some inaccuracies in the wording of the CAS provisions can lead to a mirror situation, when the participants in the proceedings themselves will disrupt the process, take it in a different direction, ”concluded Nikolai Yashin.