1T vs 2T

[zakee00]

http://forums.pimprig.com/showthread.php?t=33809

I was having trouble getting my Geil memory to perform as advertised. At 240MHz, it would fail memtest at its 250MHz timings. That was with 2.85v also. Right now, i am testing at 3-4-4-10, at 240MHz, 1T and there has been no errors so far at 800%.
The reason i found that my memory wasnt passing memtest at its rated timings was because i was running at 1T instead of 2T.
When i run at 1T 3-4-4-10, i get 7,100 in sisandra. with 2T 2.5-4-4-7 i get a mere 5,800.
im going to leave it at 2T for stability, but speed suffers alot as a result.

hope this was sort-of informative,
Nick

 

[stratman]

I'd imagine that running 1T at a lower (stable) frequency would be better than going 2T.

 

[Duvie]

Also remember according to Zebos results that may not hurt that much.....A mere 1300 in sissoft mem scores does not translate into much in the real world....Do what you have to for stability....

 

[zakee00]

ok, thanks duvie. that is your ram at btw? your system is beautiful. im guessing you do 3d work, not game? quadro fx?

actually, stratman, 1t is WAY faster then 2t. 2t is for stability.

 

[h3nG]

uh go with what's stable...

lets me see, maybe 1fps or perhaps 2fps improvement in games? or maybe your programs will load 0.1 sec faster. if you can notice those improvements and they are that important, i guess.

hardly worth the effort though...and future hassle of debugging, and/or corrupt files from memory failure.

 

[Avalon]

Yeah, if you're running on an A64 system, I would expect running at 2t wouldn't damper very much of anything, since the A64 isn't too worried about bandwidth.

 

[Duvie]

I am running cas 2.5,4,4,10, 1t with a 166divider at the 330HTT.....I need to rma a stick og Gskill that will not run their advertised cas 2.5,3,3,7 at 550ddr let alone at this 520ish ddr with cas 2.5,3,3,7....

I tested it and the ram stick will not load if in slot 1, corrupt hdd if in slot 3.....It will run fine in slot 2 if at this speed or lower.....Slot 4 will give errors...

I did a massive test of memtest one night where I literally tried all sorts of speeds, divider, vdimm, etc...even waited for my new powersupply and nothing changed....Massive errors in test #5 and #6 near rated (advertised speeds)...Mysteriously the errors would always register between 800-1000mb range...Hmm sounds like same stick in slot 2 that can't work elsewhere....

My one stick in slot 1 with limited testing did 620ddr with cas 3,4,4,8, and 590ddr at cas 2.5,4,4,8...all 1t


Yep the quadro fix of my 6800 is my crown jewel at the moment. PPL just need to realize the money saved by doing something like that...I will be looking at doing the hardmod/ bios flash so it will be permanent (not a sotfware hack that is reversed when you format) and I can run in Linux....

 

[zakee00]

yay im FINALLY prime95 stable, from switching from 1T to 2T. now ill see how much farther i can push my system
damn this gets addictive...

 

[Icopoli]

Ok so the general consensus is that 1T/2T makes no difference?

 

[GuitarDaddy]

it definately makes a difference, just not a huge difference. Definately not worth sacrificing stability for.

 

[ssvegeta1010]

Zakee, is the RAM rated for 1T at 250 Mhz? If it is, then you can return it for not running at the specified settings.

Otherwise, 1T/2T isnt a horrible difference, as said by GuitarDaddy.

 

[Duvie]

Originally posted by: GuitarDaddy
it definately makes a difference, just not a huge difference. Definately not worth sacrificing stability for.

Like in running superpi 2mb it made like 2 sec using same cas timings and ddr speed....superpi is bandwidth happy...Zebo didn't show it on his test cause the 1mb test is too short.....

IN encoding itime of a 2 min clip in TMPGenc it made 1 sec.....1 sec!!! in a 2 hour movie that may make a difference of 1 minute...whoopty Dooooooo

IN rendering a frame in my Viz renderer it took less then 1 second longer (had to use a stop watch) in rendering one frame of 1024x768...NOw with lighting and shadowing on it already took 23 seconds to process one single frame...take 23x24fpsx(15 sec clip)=2hrs and 18 minutes....A less fluid 640x480 at 15fps and using a different codec and I can cut that by a considerable amount.....Hwever figuring 1sec 2t would have cost me 6minutes or 3minutes for 1/2 a second....overall minute....


In games it seems to make more a duifference if you trust 3dmark as a definitive gaming benchs...It like bandwidth and more so ther 01-03 versions...

 

[Acanthus]

:thumbsup:

I *LOVE* playing SuperPi!

 

[Devrin]

A bit off-topic here but I just came across you talking about the 6800 to FX4000 mod. I do quite a bit of cad work as Im a engineering student, but in my free time I also like to game quite a bit. I'm in the process of upgrading to a A64(waiting on Venice) and later I plan to upgrade my gfx card, so this caught my eye and I just got a few questions if you don't mind.

Im just curious on how does gaming go when using the software hack to change the 6800 to FX4000? I know workstation cards suck for gaming, but with a softmod is this the case or is there an easy way to switch it back so its a 6800 again? Even if you can just offer some links for me to read up on this it would be appreciated, I don't mind reading long tech posts

Now back on the topic of 1T & 2T. I know when filling all 4 dimms with a winchester core it kicks you to 2T, but also drops your speed to 333mhz. Now that this is suppose to be somewhat fixed with the Venice core, it will only kick you to 2T but maintain 400mhz, would this really affect performance that much? I'm a bit curious about this as I plan on starting with 1gb with the A64 setup, if I later want to move up to 2gb would I be better off replacing the 2x512mb sticks with 2x1gb, or could I just add 2 more sticks for a total of 4x512mb? I know ideally the 2x1gb would probably be ideal but as I said I'm a student so the price/performance equation is always looming in my wallet. T

 

[Avalon]

Originally posted by: Avalon
Yeah, if you're running on an A64 system, I would expect running at 2t wouldn't damper very much of anything, since the A64 isn't too worried about bandwidth.

I'm going to have to correct my 3 month old (lol) statement. Running at a 2T command rate is pretty brutal. I was using it temporarily when I first got my sempron 2800+ to try and figure out why I couldn't get past 265HTT with my Epox motherboard.

Anyway, it took a noticeable hit on my gaming performance. I'm not sure if I remarked on how much of a hit I took in my sempy thread. I think I did. If I recall, it took me a 400mhz overclock on 2T to make up for the loss in FPS in Doom 3, FarCry, and HL2.

 

[andy nowotny]

2T will be more stable