Please report all spam threads, posts and suspicious members. We receive spam notifications and will take immediate action!
Page 1 of 4 123 ... LastLast
Results 1 to 10 of 37

Thread: tREAD and CAS Latency?




  1. #1
    Join Date
    Jun 2009
    Posts
    57

    Default tREAD and CAS Latency?

    Is there some sort of direct relationship between these two settings?

    Tonight, I was trying and trying quite unsuccessfully to get a 9.0x450@2.00D config stable. No matter what combinations of Vcc, Vtt, PLL, MCH, MCHRef, DRAMv I tried, nothing made it past 7min. in p95 Large FFT.

    After exhausting all likely voltage possibilities, I figured it was tREAD. So, I upped it from 8 to 9, and then 9 to 10; and re-tried a bunch of voltages - and still no luck.

    This was all tested with 4-4-4-12 Standard DRAM Timings. (I achieved stability with these timings for a 9.5x422@2.00D config, so I figured I might as well try them here...)

    Finally, I resorted to try looser DRAM timings. I set them to 5-5-5-15, and kept tREAD=8; and the system would not POST no matter what my voltages were set to. Then I set tREAD=10, and it POSTed booted to Windows fine.

    Then as a test, I set CAS latency back to 4, but kept the other settings: 4-5-5-15; and set tREAD back to 8; and it booted fine.

    So, my question is:
    Is there some sort of relationship between CAS Latency and tREAD? Like if CAS Latency=5, then tREAD must = 10 (9?) or higher? Or something like that? Maybe it's also somewhat dependent upon the FSB value? Or MCHv?

    And if that is true, is it advantageous to keep CAS Latency low, even if other Standard Timings are looser; so that a lower tREAD setting may be used?

    PS I'm stability-testing the 9.0x450@2.00D config in p95 now with those 4-5-5-15 timings and tREAD=8, and things are looking good, now into the 3-hr. mark! [Update: still running strong @ 10hrs.; as of this morning...]
    Last edited by corlay; 07-15-2009 at 09:25 PM.
    case: Gigabyte 'Triton 180'
    power: Corsair 'HX-620'
    motherboard: Gigabyte 'GA-EP45-UD3P' (rev.1.1, F9)
    cpu: Intel 'e8500' (E0)
    cooling: Xigmatek 'HDT-S1283'
    memory: OCZ 'Reaper' 4x2gb (OCZ2RPR10664GK)
    storage: Western Digital 'Black' (500gb)
    video: EVGA 'GeForce' GTX-260
    display: Samsung 'SyncMaster' 2333sw

  2. #2
    profJim's Avatar
    profJim is online now Chief Munchkin + moderator
    Join Date
    Dec 2008
    Location
    Tacoma, WA. [USA]
    Posts
    7,811

    Default Re: tREAD and CAS Latency?

    With 4x2GB memory, running tighter memory timings will be harder than with 2x2GB.

    The following applies to running with 4GB total memory:
    With tCas = 4, memory speeds will usually top out in the 875 - 925MHz range.
    Using the 2.00B memory multiplier, tREAD = 8 is optimal in the lower to mid-range 400MHz cpu settings. 2.00D will need to be one click higher in this range compared with using 2.00B. tRead = 7 is usable up to about 425MHz. With 8GB of ram, these ranges might have to be somewhat lower. With your 9.5x422 settings, your memory was running at 844MHz.

    My 7.0 x 459 (2.40B) configuration is stable using tRead = 8 with 5-5-5-15 memory timings. Memory speed was 1102MHz with my 2x2GB Mushkin memory. My current setup (see my signature) is noticeably faster using 12% lower memory speeds and 7% higher cpu speeds.

    Some general guidelines:
    With 8GB, MCH Core will need to be higher than the same memory with 4GB.
    tRead settings are dependent on your Cpu Host Frequency.
    Primary memory timings are dependent on overall memory speed.
    It all boils down to finding the best balance of cpu and memory speeds with the appropriate stable settings.

    Running Prime95 with Large FFTs will stress your memory much more than using the Blend or Small FFT tests.
    Q9650 @ 4.10GHz [9x456MHz]
    P35-DS4 [rev: 2.0] ~ Bios: F14
    4x2GB OCZ Reaper PC2-8500 1094MHz @5-5-5-15
    MSI N460GTX Hawk Talon Attack (1GB) video card <---- SLI ---->
    Seasonic SS-660XP2 80 Plus Platinum psu (660w)
    WD Caviar Black WD6401AALS 640GB (data)
    Samsung 840 Pro 256GB SSD (boot)
    SLI @ 16/4 works when running HyperSLI
    Cooler Master 120XL Seidon push/pull AIO cpu water cooling
    Cooler Master HAF XB computer case (RC-902XB-KKN1)
    Asus VH242H 24" monitor [1920x1080]
    MSI N460GTX Hawk (1GB) video card
    Logitech Z-5500 Digital 5.1 Speakers
    win7 x64 sp1 Home Premium
    HT|Omega Claro plus+ sound card
    CyberPower CP1500PFCLCD UPS
    E6300 (R0) @ 3.504GHz [8x438MHz] ~~ P35-DS3L [rev: 1.0] ~ Bios: F9 ~~ 4x2GB Kingston HyperX T1 PC2-8500, 876MHz @4-4-4-10
    Seasonic X650 80+ gold psu (650w) ~~ Xigmatek Balder HDT 1283 cpu cooler ~~ Cooler Master CM 690 case (RC-690-KKN1-GP)
    Samsung 830 128GB SSD MZ-7PC128B/WW (boot) ~~ WD Caviar Black WD6401AALS 640GB (data) ~~ ZM-MFC2 fan controller
    HT|Omega Striker 7.1 sound card ~~ Asus VH242H monitor [1920x1080] ~~ Logitech Z-5500 Digital 5.1 Speakers
    win7 x64 sp1 Home Premium ~~ CyberPower CP1500PFCLCD U.P.S
    .


  3. #3
    Join Date
    Jun 2009
    Posts
    57

    Default Re: tREAD and CAS Latency?

    Quote Originally Posted by profJim View Post
    The following applies to running with 4GB total memory:
    With tCas = 4, memory speeds will usually top out in the 875 - 925MHz range.
    Thanks Jim, that's good info to know.
    After tightening-down my 9.0x450@2.00D config,
    I had planned to go for something like 9.0x467@2.00D next -which puts my above the ~925Mhz range.

    Question: When treading upon the cusp of stability, with regard to FSB and tRD, is my solution of keeping tRD low, and loosening the remaining Standard Timings a proven way to keep DRAM performance high?

    Using the 2.00B memory multiplier, tREAD = 8 is optimal in the lower to mid-range 400MHz cpu settings. 2.00D will need to be one click higher in this range compared with using 2.00B.
    Thanks, again. I will use tRD=8 as the goal for my future tweaks, assuming that I'm FSB > 450.

    tRead = 7 is usable up to about 425MHz. With 8GB of ram, these ranges might have to be somewhat lower. With your 9.5x422 settings, your memory was running at 844MHz.
    yes, this makes perfect sense. I was able to use tRD=7 with 4-4-4-12 timings for a 9.5x400@2.00D config, but had to jump tRD to 8 for 9.5x422@2.00D. So, I've experienced exactly what you've suggested (my 8gb lowering the bar a bit below the 425MHz threshold...)

    My current setup (see my signature) is noticeably faster using 12% lower memory speeds and 7% higher cpu speeds.
    This is due to your running of your DRAM at 2.00(B/D) (1:1 ratio) in your current config, vs. the 2.40B in your previous one, correct?

    Thanks for this info. This is one aspect of overclocking that I'm still trying to learn: What is truly important to focus on with regard to overall performance, and what is not? Like, should I really be fretting over reducing my Advanced DRAM Timings to the very brink of instability, or is that really just a waste of time?

    Some general guidelines:
    With 8GB, MCH Core will need to be higher than the same memory with 4GB.
    tRead settings are dependent on your Cpu Host Frequency.
    Primary memory timings are dependent on overall memory speed.
    It all boils down to finding the best balance of cpu and memory speeds with the appropriate stable settings.
    words to live by. Thanks.

    Running Prime95 with Large FFTs will stress your memory much more than using the Blend or Small FFT tests.
    I had thought that "blended" just alternated between small and large FFT's? What does running a dedicated Large FFT test tell you that a blended test doesn't?

    Thanks again for all of your advice.
    Very helpful...
    Last edited by corlay; 07-15-2009 at 10:11 PM.
    case: Gigabyte 'Triton 180'
    power: Corsair 'HX-620'
    motherboard: Gigabyte 'GA-EP45-UD3P' (rev.1.1, F9)
    cpu: Intel 'e8500' (E0)
    cooling: Xigmatek 'HDT-S1283'
    memory: OCZ 'Reaper' 4x2gb (OCZ2RPR10664GK)
    storage: Western Digital 'Black' (500gb)
    video: EVGA 'GeForce' GTX-260
    display: Samsung 'SyncMaster' 2333sw

  4. #4
    Chike is offline Senior Member
    Join Date
    Jun 2009
    Location
    Israel
    Posts
    6,531

    Default Re: tREAD and CAS Latency?

    Quote Originally Posted by corlay View Post
    I had thought that "blended" just alternated between small and large FFT's? What does running a dedicated Large FFT test tell you that a blended test doesn't?
    The difference is the in place option. Not sure what is is, only a guess is it uses the same memory during the test and not allocate different blocks of memory.
    With small FFT all data will be in the CPU cache and memory will not be accessed, and large will cause to be rapidly commited abd loaded when cache lines collide.

    Selecting blend and then custom, cheking the in place box will be as running small and large FFT blend test.

  5. #5
    Join Date
    Jun 2009
    Posts
    57

    Default Re: tREAD and CAS Latency?

    Quote Originally Posted by Chike View Post
    The difference is the in place option.
    from the whatsnew.txt of the Prime95 installation files:

    There are now 3 flavors of torture testing. One that operates on small FFTs only, theoretically maximizing FPU stress. One that operates in place on large FFTs only, maximizing heat and power drain. One that runs a blend of tests using lots of memory, possibly catching errors that might only occur at a specific memory address.
    After reading this, it seemed like *not* selecting "in-place" for the Blended Test would do a better job of catching Memory errors? But, re-reading it again, maybe "in-place" stresses the system more fully with regard to voltage levels?
    case: Gigabyte 'Triton 180'
    power: Corsair 'HX-620'
    motherboard: Gigabyte 'GA-EP45-UD3P' (rev.1.1, F9)
    cpu: Intel 'e8500' (E0)
    cooling: Xigmatek 'HDT-S1283'
    memory: OCZ 'Reaper' 4x2gb (OCZ2RPR10664GK)
    storage: Western Digital 'Black' (500gb)
    video: EVGA 'GeForce' GTX-260
    display: Samsung 'SyncMaster' 2333sw

  6. #6
    Chike is offline Senior Member
    Join Date
    Jun 2009
    Location
    Israel
    Posts
    6,531

    Default Re: tREAD and CAS Latency?

    Quote Originally Posted by corlay View Post
    After reading this, it seemed like *not* selecting "in-place" for the Blended Test would do a better job of catching Memory errors? But, re-reading it again, maybe "in-place" stresses the system more fully with regard to voltage levels?
    If you are looking to test memory faults then the blend is the test to run, if you want to test memory settings then large in-place is likely to find them faster.
    Last edited by Chike; 07-15-2009 at 11:14 PM.

  7. #7
    profJim's Avatar
    profJim is online now Chief Munchkin + moderator
    Join Date
    Dec 2008
    Location
    Tacoma, WA. [USA]
    Posts
    7,811

    Default Re: tREAD and CAS Latency?

    Keep in mind that the info in post #2 are general guidelines, not necessarily hard and fast "rules".

    My approach is to overclock with looser memory timings with an "appropriate" tRD (Static tRead Value) setting. If it passes IBT and Prime95 Blend settings, then I'll tighten the memory timings and run additional stress/stability tests using Large FFTs to concentrate on memory. OCCT (v3.1) is also an excellent tool that performs similar tests and has the added benefit of saving graphs that include various voltages and temperature readings. Others have found that extended Prime95 or OCCT tests will run with no errors for 12 -24 hours, only to fail after 6 - 10 hours when testing with Large FFTs.

    Memory performance is mostly related to tRD and primary memory timings. tCas and tRD seem to have the most effect on memory performance. My approach is to set tRD to the lowest stable/stress setting and then focus on the primary memory settings. This is how I found that 5-5-4-12 settings were the tightest I could run on my system with the settings in my signature. I don't rely on synthetic benchmarks and prefer to test using repeatable real world tests including encoding, image processing, virus scanning, or game playing to judge what works best. You might find that the best overall performance is with running with memory speeds in the 850 - 900MHz range with tighter timings.

    I've found that the 2.00B multiplier let's me use a lower tRD setting up to about 460MHz cpu frequencies. At higher cpu host frequencies, the 2.00D multiplier runs with a lower tRD setting than the 2.00B multiplier. Some have found that the 2.00D multiplier is easier to get stable than the 2.00B multiplier.

    Tweaking Advanced Memory Timings should be left to last, as they are very time consuming. Once I settled on my current setup, I was finally able to run with stable 1-3-3-48-1 settings. Your setup will probably be stable using tRFC = 60 - 64 with 8GB of memory. I'd leave the other advanced settings on AUTO for the time being. After you've reached a stable setup, setting Performance Enhance = Turbo might improve performance somewhat, but you'll need to repeat all of your stability tests.
    Q9650 @ 4.10GHz [9x456MHz]
    P35-DS4 [rev: 2.0] ~ Bios: F14
    4x2GB OCZ Reaper PC2-8500 1094MHz @5-5-5-15
    MSI N460GTX Hawk Talon Attack (1GB) video card <---- SLI ---->
    Seasonic SS-660XP2 80 Plus Platinum psu (660w)
    WD Caviar Black WD6401AALS 640GB (data)
    Samsung 840 Pro 256GB SSD (boot)
    SLI @ 16/4 works when running HyperSLI
    Cooler Master 120XL Seidon push/pull AIO cpu water cooling
    Cooler Master HAF XB computer case (RC-902XB-KKN1)
    Asus VH242H 24" monitor [1920x1080]
    MSI N460GTX Hawk (1GB) video card
    Logitech Z-5500 Digital 5.1 Speakers
    win7 x64 sp1 Home Premium
    HT|Omega Claro plus+ sound card
    CyberPower CP1500PFCLCD UPS
    E6300 (R0) @ 3.504GHz [8x438MHz] ~~ P35-DS3L [rev: 1.0] ~ Bios: F9 ~~ 4x2GB Kingston HyperX T1 PC2-8500, 876MHz @4-4-4-10
    Seasonic X650 80+ gold psu (650w) ~~ Xigmatek Balder HDT 1283 cpu cooler ~~ Cooler Master CM 690 case (RC-690-KKN1-GP)
    Samsung 830 128GB SSD MZ-7PC128B/WW (boot) ~~ WD Caviar Black WD6401AALS 640GB (data) ~~ ZM-MFC2 fan controller
    HT|Omega Striker 7.1 sound card ~~ Asus VH242H monitor [1920x1080] ~~ Logitech Z-5500 Digital 5.1 Speakers
    win7 x64 sp1 Home Premium ~~ CyberPower CP1500PFCLCD U.P.S
    .


  8. #8
    Join Date
    Jun 2009
    Posts
    57

    Default Re: tREAD and CAS Latency?

    Quote Originally Posted by profJim View Post
    I've found that the 2.00B multiplier let's me use a lower tRD setting up to about 460MHz cpu frequencies.
    I've had *zero* luck on the B-strap with my system.

    Your setup will probably be stable using tRFC = 60 - 64 with 8GB of memory. I'd leave the other advanced settings on AUTO for the time being.
    I've been testing with 4-4-6-54-4-2 manually set Advanced Timings, and have not run into any issues so far. For my 9.5x422@2.00D config I successfully squeezed them down to 3-4-4-48-4-2; and didn't attempt to go any lower than that. (you're right, it does take up a lot of time!)

    After you've reached a stable setup, setting Performance Enhance = Turbo might improve performance somewhat, but you'll need to repeat all of your stability tests.
    I'll keep this in mind.
    Thanks!
    case: Gigabyte 'Triton 180'
    power: Corsair 'HX-620'
    motherboard: Gigabyte 'GA-EP45-UD3P' (rev.1.1, F9)
    cpu: Intel 'e8500' (E0)
    cooling: Xigmatek 'HDT-S1283'
    memory: OCZ 'Reaper' 4x2gb (OCZ2RPR10664GK)
    storage: Western Digital 'Black' (500gb)
    video: EVGA 'GeForce' GTX-260
    display: Samsung 'SyncMaster' 2333sw

  9. #9

    Default Re: tREAD and CAS Latency?

    The allowable tRD setting depends on 3 settings: CAS latency, FSB, and the DRAM to FSB ratio. Low CAS latency, low FSB and high DRAM to FSB ratio favor lower tRD settings.

    A good read on the subject: AnandTech: ASUS ROG Rampage Formula: Why we were wrong about the Intel X48.
    QX9650 batch L739A761/ GA-EP45-UD3P/ Kingston KHX9200 4x1G

  10. #10

    Default Re: tREAD and CAS Latency?

    Oh that's a great article, freecableguy wrote that article, too bad he doesn't do much writing anymore. He wrote some awesome in depth articles.
    He kind of disappeared from the scene about the time DDR3 was being introduced.
    Most of his articles deal with DDR2, and can be found on the Techrepository

    TTR's Guide to Determining Optimum Overclocked System Performance Points - The Tech Repository Forums
    Mushkin Enhanced comes alive...

    Need help with your Mushkin Product? Click Here

Page 1 of 4 123 ... LastLast

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •