Has Anyone seen a 1T vs. 2T command rate thread for A64's?

[ImageRoc]

I encoded a movie using NeroVisionExpress using all the same settings except once with 1t then 2t
both times the movie took 15mins, 1T was just a few seconds quicker.

The only thing I do with my rig, right now is encode movies this proves for me my real world applications that I use have no gain with 1T settings

 

[Hummercash]

anyone here have 3700+ sd benchmarks? i have one of those and 2gb (two sets of 2x512mb ocz el rev2 3200) of ram... but no other parts yet. gonna be placing my order for the rest of my system tmrw, but would love to know if i should keep all the ram or sell off 1gb of it.


.//chris

 

[Goi]

With 1T I'm barely able to get my Newcastle 3200+ running above 2.2GHz. With 2T, I'm able to run 10x250MHz on my Kingmax DDR500 modules no problemo.

 

[Lord Banshee]

Thats strange..

My 2800+ Newcastle works at 2300 with 1T or 2T.... My overclock doesn't even go up with more VDimm or looser timming.

For me SuperPi experience a 5% increase with 1T than 2T I belieave. I'll do a benchmark when i get home tonight.

 

[eBauer]

Only had time to run through 3dmark 2001. The individual tests tell more of a story as opposed to posting just the final score.

754 Sempron 2800+ @ 2.4GHz (300x3 HTT), RAM @ /12, 200MHz 2-2-2-10, 9800P @ 412/360
Stock test @ 1024*768

Car Chase Low 1T - 290.4
Car Chase Low 2T - 285.2
1T is 1.8% faster

Car Chase High 1T - 108.3
Car Chase High 2T - 103.2
1T is 4.7% faster

Dragothic Low 1T - 390.9
Dragothic Low 2T - 387.2
1T is 0.9% faster

Dragothic High 1T - 212.7
Dragothic High 2T - 210.5
1T is 1.0% faster

Lobby Low 1T - 281.8
Lobby Low 2T - 267.0
1T is 5.3% faster

Lobby High 1T - 126.4
Lobby High 2T - 118.9
1T is 5.9% faster

Lobby has typically responded well to memory bandwidth so there is little surprise it benefits from 1T more than the other tests. Unless you're obsessed with the ORB, 2T will work just fine

 

[Holy Smoke]

3DMark2001 @ 640x480x32

Winchester 3200 @ 2.2.ghz 9x250FSB DDR400 @ 3-4-4-8 6600GT @ 500/500

Car Chase Low 1T - 305.4
Car Chase Low 2T - 301.2
1T is 1.4% faster

Car Chase High 1T - 102.0
Car Chase High 2T - 95.4
1T is 6.9% faster

Dragothic Low 1T - 415.5
Dragothic Low 2T - 402.1
1T is 3.3% faster

Dragothic High 1T - 213.2
Dragothic High 2T - 206.4
1T is 3.3% faster

Lobby Low 1T - 281.3
Lobby Low 2T - 268.9
1T is 4.6% faster

Lobby High 1T - 125.8
Lobby High 2T - 120.1
1T is 4.7% faster

Note that this is at 640x480 since I wanted to remove the graphics card as far out of the equation as possible.

 

[caboob]

This is great work guys. i now run my 2GB of BH-UTT at 240 at 2T as I get more stable operation.

 

[Gamingphreek]

Anyone have any results for an AXP or P4 based system.

-Kevin

 

[rise]

winnie @ 2475 | 2.5/4/3/7 @ 275

1t -- 2t
super pi 1.04 8M __ 6:10 ___ 6:27
pcmark04 __ 5143 ___ 4996
aquamark __66,903 ___ 65,615
3dmark01 __24,361 ___ 23,212
3dmark03 __11,710 ___ 11,592

 

[Leper Messiah]

will do some tests when I get home tonight, as I just picked up another stick of PQI. right now running 2.5-3-3-7, 2T, 3dm01 in the low 24k range, from what I remember. Sandra gets about 5200MB bandwidth dual channel, more than enough to outweigh my 1T single timming scores...

 

[Leper Messiah]

My pre-lim results, just sandra and 3dmark01. some of them are a little weird, I can't explain. I did this at 11:50 last night, after working 13 hrs, so give me a break!
Sandra Scores 233 3-3-3-8 1T 6751/6683MB.
Sandra Scores 233 3-3-3-8 2T 5532/5532MB. 18% difference in scores

3d mark 01 3-3-3-8 1T 23816
car chase low 314.8
car chase high 118.4
Dragothic low 425.0
Dragothic high 235.9
Lobby low 324.9
Lobby High 147.7
Nature 143
3d mark 01 3-3-3-8 2T 23181 -2.7%
car chase low 318.3 +1.1%
car chase high 116.2 -1.9%
Dragothic low 422.4 -0.7%
Dragothic high 232.5 -1.5%
lobby low 312.5 -3.9%
Lobby high 140.9 -4.7%
Nature 142.9 neg.

3d mark 01 2.5-3-3-7 2T 23444 -1.6%
car chase low 335.1 +6.4%
car chase high 116.4 -1.7%
Dragothic low 422.5 -0.6%
Dragothic high 233.6 -1.0%
lobby low 315.3 -3.0%
lobby high 142.9 -3.3%
Nature 142.9 neg.

 

[AkumaX]

My pre-lim results, just sandra and 3dmark01. some of them are a little weird, I can't explain. I did this at 5:50pm, after working 8 hrs, so give me a break!

Super Pi 8mb 7:03m 1T
Super Pi 8mb 7:22m 2T (4.3% difference)

Sandra Scores CPU Arithmetic Bench -- The ALU was the same, but the FPU/iSSE2 was 5242(1T) vs 4876(2T) for 7% difference
Sandra Scores Memory Bench 1T 3007/3006MB Buffered -- 2598/2594 Unbuffered.
Sandra Scores Memory Bench 2T 2690/2691MB Buffered -- 2239/2225 Unbuffered. (11% difference Buf, 14% difference Unbuff)

Everest (AIDA32)
Memory Read 3029 1T -- 2981 2T (1.6% difference)
Memory Write 1326 1T -- 1086 2T (18.1% difference
Memory Latency 51.9 1T -- 58.2 2T (10.8% difference)

3DMark01 1T 19522
3DMark01 2T 18842 (4.5% difference)

3DMark03 1T 8740
3DMark03 2T 8671 (0.8% difference)

edit: oh yeah, i'm different and am using single channel solution == socket 754

Athlon 64 3200+ 10x240(2400 mhz) 1.55vcore
ECS 755-A2 SIS 755
2 x 512mb Corsair VS 2.5-3-3-7 @ 200mhz
BFG 6600GT OC @ 550/1.1ghz
160gb Maxtor DM9 8mb PATA

 

[Duvie]

Synthetic ppl!!!! Run real world games and test...I would soon as wipe my arse with sissoft anything.....

Badnwidth intensive apps like games seem to be in the 2-4% range.....winrar which we all know loves bandwidth can be in the 7% range.....cpu intensive apps like encoding (audio and video) 1% or less in my testing.....

I wouldn't hesitate to use it if I thought I could get another 50-100mhz out of it....Unfortunately for me I tried and I am at the end of the cpu....

 

[hippotautamus]

I agree with Duvie, synthetic benchmarks are for stroking the e-penises of the impotent! Let's see some FPS and render times in here, lads.
What we really need is a new high end RTS (AOEIII!!!) to test with Fraps. If any type of game is a memory whooore it's RTS games. When I went from 256mb of pc133 to 1gb of DDr400, I saw the most ridiculous jump in RTS goodness...I could have an 8 player battle in Age of Mythology at 1600x1200 with all graphic detail and pop limits maxed out with absolutely zero lag, I thought I had died and gone to (aom)heaven.

 

[Leper Messiah]

^^^buy a new GPU so that I'm not CPU limited in anything you guys care about. Buy me some games that I can test with, and you'll get your results. I use what I have.

 

[Duvie]

Who sia djust games.....There are other apps I am sure you use that dont need the gpu...synthetic test are goig to give you skewed results and as seen percentages waaaaaaaaayyyyyy above actual real world.....Proably why the "doom and gloom" ppl as Zebo suggested blasted 2t as being the death nail....I never bother with anyone running synthetics.....Just isn't real.....

 

[Leper Messiah]

I don't really video edit. I encode MP3's but thats about it. 1t/2T isn't going to make a seconds difference in encoding speed. other than that, I game, and I write papers, and surf the net.

 

[rise]

i kinda thought this was dead but i ran couple other times the other day--

far cry volcano 12x10, maxed settings
1t 89.71
2t 87.65

dvdshrink encoding no compression
1t 15.40
2t 15.40

dvdshrink encoding compression 10:6
1t 28:40
2t 29.42

 

[MelonJuice]

Hi, I'm new here... I just recently search for 1T/2T command rate information, and It's brought me here...
I have 2 pair of Corsair TWINX1024-3200LL running in 2-3-2-6 @ 2.7v , which bring me to A64 memory problems, because I'm running 4 x 512mb rams.

My current setup:
AMD64 3500+ Socket 939 "Clawhammer"
Gigabyte K8NS-ULTRA 939
2gb Corsair CMX512-3200LL x 4

The problems start when the first time my system arrived.
I'm trying to stick in my old memory into this NF3 motherboard (2 x 512mb), then the first error comes up which resulting re-booting when installing WinXP pro.
I try to figure out what's the problem? and my friend lend me a pair of 512mb generic rams... so the current setup works - no error encountered even crash or hang. (4 x 512mb).
After I gave it back her rams, It bring me to an idea to get a match pair for my old rams - then I search ebay (Because it has been discontinued) - and I succesfully to get one.

Second problems struck me.
More crash, more hangs, this totally makes me headache! I can't find any solution to make them working without errors.
And I ask RamGuy in Corsair official forum, but one of the member suggest me to downgrade the speed to DDR333@166mhz - It's work, however I dont want to end this all.
I have spend so many cash on this rams, so I keep trying to find good solution to run this rams in their native speed (DDR400@200mhz).

Until I finally want to give up,
but someone in other forum talk about 1T/2T command rate, and I give it a try to my rams and finally its WORK! (Default setting set to 1T)
I run memtest86 with both pair TWINX1024-3200LL under 2T/200mhz/2-3-2-6/2.7v - no errors in test (for 8 hours each)
But! when running 4x512mb under 2T/200mhz/2-3-2-6/2.7v - 16 errors in Test no.7 (12 hours test x 2 times)
However, I still run it till now - no hang/freeze or whatsoever encountered. - but still my anxiousity/curiousity bring me to this thread.

Today I run, Sisoft Sandra 2005 - under standard memory benchmark it resulted only 4336mb/sec (which is surprisingly low for 2gb?)
I haven't make comparisson for 1T/2T because my rams unable to run 1T command rate (Because the ram is double bank?)
anyone have suggestion how to fix this problem? (I will sell the other pair TWINX1024, and I will back to original state in 1gb, but I still need 2gb because I do much 3D stuff and video editing)

 

[MelonJuice]

bump!

 

[Zebo]

First set Tras to 8. put all four dimms in.

You fail memtest86? try tests 5 and 6 about 3 times each...

If that fails try increasing Vdimm to 2.8.

If that fails still lower vdimm back to 2.7 and try 2.5-3-2-8 2T



No way you can run 1T with 4x512, only ram that can do this is crucial 8T.

 

[Towndog]

I'm having probs finding anywhere to change to 1T for testing on my MSI NEO Plat 754 board. Running 1.6 BIOS & I only have max memclock, CAS,RAS to CAS,Tras,and Trp settings available. Help please.

 

[MelonJuice]

Hi, thnx for the response!
Here my latest report...

1st test - 2-3-2-8/200mhz/normal voltage(does this set to 2.5 or 2.7 in gigabyte motherboard?) = Test 7/ 5 errors.
2nd test 2-3-2-8/200mhz/+.02v = Test 7/ 1 error

now I'm running in 2.5 3-2-8 under Normal voltage, all in 2T.
but haven't memtest it yet, coz i sick looking blue screen
i'll do it tommorow... but any suggestion?

 

[zendari]

So whats the bottom line? Is the 1T vs 2T thing real or rumor?

 

[imported_Starman]

Originally posted by: zendari
So whats the bottom line? Is the 1T vs 2T thing real or rumor?

It is real, but it appears the difference only varies from 1-5%. However, it's important to remember that this is just one aspect of achieving the best performance for your PC. A few percent here, a few percent there, and all those percentage points quickly add up. Most people who want to see a negligible difference between 1T vs. 2T are also those who want to see negligible differences between low-latency RAM and Value stuff. What goes so often forgotten is that all these factors eventually add up and you're no longer talking about a negligible difference. Value RAM, 2T command rate, buying a slower processor...you've got to pick your poison and be careful not to drop the entire context of system performance variables. Before you know it you're no longer talking about 1-5% but a cumulative 10-15% or worse.