GT underperforming?

K404

New member
Hey! I`ve been benching my rig at 2925 on the CPU (243 on the RAM) with my GT at 507MHz on the core with the shader and ROP clocks at 486MHz and the RAM at 1300. This is getting me around 17779 in 03 and 8158 on 05.

No offense to ov3rclock3r here, but his scores are around 1K and 600 points above mine, despite lower frequencies. This has been posted with blessing and this aint a "boo-hoo" on my part as much as a "why"?

Hopefully ov3rclock3r can add more details to his spec but there it is.

Any ideas?

K
 
No probs.

Ok people, Me and kenny have been discussing this on msn aswell but we cant figure out the problem.

Why is it that he's scores are so low seeing as his card is volt modded, running at higher speeds with tweaks (rop stuff)?

My specs and compare links are in sig:)
 
Damn obvious i know, but are you running the same drivers? Only thing i can really think of atm, i know both of you would have clean installs etc.

Maybe your computer just doesnt like you Kenny :p

G
 
I think Master G has hit the nail on the head here - it's obvious your computer dislikes you kenny.

Seriously though this is a bit of a puzzler, I wouldn't think drivers would make as much difference as that? Unless its a power supply issue?
 
Heh yup, same drivers. I was running an older version but updated sharpish when I saw his scores! My install could be slightly cleaner, but was still stripped for the benching

I tweaked my graphics RAM settings...was thinking they might be too tight, causing corruption and the frames are having to be re-rendered?

Cant see it being power- my GPU has 20A on a 12V rail all to itself. Wouldnt that limit clocks more than reduce performance at existing clock speeds?
 
he's correct. Basically, even before i formatted, I was running a messed up version of windows but my scores were only 100-200 points lower but still higher than kenny's

Kenny, Try defaulting the timings on your card, see if that helps
 
Yea, will hook up a floppy tomorrow/ today and put the timings back.

:D I cant see that making up a K on `03 though.

All ideas welcome!

:D

K
 
Well...made up my scores to 181xx and 8500....getting closer but somethings still not quite right. Will try a clean install maybe and RAM timings :)
 
Ram timings, tweaks in all the right places and a fresh install or use something like Nlite to strip out what you have now.

Fully defragged with swap file on differnt drive to the OS - using Rivatuner to drop LOD to 15 (furthest right) and BIOS flash your clocks instead of using Riva to achieve them - ensure that 2d is as high as it can go also - it could just be as simple as his cards run at just the rights temps for optimum performance.

Mav
 
Yea, now someone mentions i presume thats how i got my number 2 score in 03.. i tightened the RAM timings on my 9800xt and upped the clock speeds, and i got a huge increase. Be careful though.. wouldnt want you breaking anything kenny!
 
Try runing each card on its own,they should score roughly the same.At least you will be able to tell if one's knackered or both are working correctly
 
I think the difference in scores are down to Ov3rclock3rs running at 545-ish core. Because he was so insistent that his clock speed was 505 I thought his was just performing extra well, but Galaxy cards have a 40 delta flashed onto them.

:)

Kenny
 
You will also find that your max overclock will not allways get your best scores without volt mods.If you overclock them to much your scores will start to come down,it's just a case of finding what set up works best as every system if different which takes time.
 
Glad you sorted it out Kenny - just to go over this point with galaxy cards - when you set the core clock in rivatuner its actually 40mhz higher? If thats the case I've got a pretty good clocker.
 
If you open up Riva and get the realtime graphs of clock speeds up in front of ya, those will show ya the real clock speeds. :)
 
name='nick25' said:
Glad you sorted it out Kenny - just to go over this point with galaxy cards - when you set the core clock in rivatuner its actually 40mhz higher? If thats the case I've got a pretty good clocker.

Dont get confused with GT's and GTX's - but the delta is put in for different parts of the core clock speed (all GTX's have a Delta clock in as Standard, not all GT's do) - I dont fully understand it but the Delta helps with different shader pipe speeds (pixel versus vertex) and then does something to the ROP - been covered before here's a linky http://www.extremeprometeia.com/forum/showthread.php?t=3349&highlight=7800
 
Not trying to go way off topic here but I remeber that link Mav and found it all a bit confusing :$ also from the few tests I did using 3dmark I didn't really find any difference when running primary clocks. Anyhow I'm happy with how things are running atm :whistle:
 
name='nick25' said:
Not trying to go way off topic here but I remeber that link Mav and found it all a bit confusing :$ also from the few tests I did using 3dmark I didn't really find any difference when running primary clocks. Anyhow I'm happy with how things are running atm :whistle:

Thats great Nick mate - just dont think that your GTX (set at the clocks you set it at) is running faster than anyone elses set at the same speed - some of the clock does run at that speed, but 3dmark for example will display your clock speed plus the Delta, really it is only reporting the fastest part of the clock.

Any yes, it is confusing, although setting a delta to 50 apparently does help with clocks and stuff for benching - not noticeable for a non-bencher like yourself really :whistle: .

Mav
 
Kenny, my card is running 505 core but it has the 40 delta on it, Thats why it shows the core as being run at 545mhz.

Anyways, I've modded my bios to 1.5v, 50 delta and will be flashing it soon.:D
 
Back
Top