512MB HIS X1800XT PCI-E + AMD64 Opteron 170 Dual Core adventures

[eva2000]

With ATI Catalyst 5.12 being released and optimised for dual core cpus, I decided to swap my 512MB HIS X1800XT PCI-E card over to my Opteron 170 dual core cpu rig

You can re-read the 512MB HIS X1800XT single core results with Opteron 146 at http://www.xtremesystems.org/forums/...ad.php?t=81585



System

Opteron 170 CCBWE 0544XPMW cooled by Corsair Hydrocool 200EX
DFI NF4 UT SLI-DR R.AB0 11/14-3 bios
512MB HIS X1800XT with Cat 5.12 drivers
4x 80GB Hitachi 7K80 SATAII Raid 0 OS drive
1x 250GB Maxtor Maxline Plus II PATA
Pioneer DVD Burner
Floppy drive
650W Silverstone SLI PSU

WinXP Pro SP2
NF4 6.53WHQL chipset drivers

Using 2x 1GB OCZ PC4000EB Platinum 3-3-2-8 rated memory
@3-3-2-5 7-14-2223 at 2.64v vdimm

===================================
Overclocking X1800XT
I'm overclocking and changing voltages on X1800XT the same 5 step method i used on single core Opteron 146:

1. In service manage in windows disable ATI Smart and ATI Hotkey services and reboot pc. You loose temp reading in CCC but can use ATI Tool to view temps.
2. Use X1800 Oc'ing tool and set 3D desired clocks using the 2D panel
3. Use ATI Tool to only verify and view the 3D clocks are set correctly. So should be same as what you set in X1800 Oc'ing tool.
4. Whenever you set the mem clock, remember to set the GPU core clock again, or it will fall back to lower clocks and when you set the GPU core clocks again, remember to set the mem clock again, or the same will happen.
5. Setting voltages, OC'ing tool reports wrong voltages so DO NOT use OC'ing tool to set voltages, use ATI Tool to set voltages. Example, GPU voltage reported in ATI Tool is 1.275v will show in OC'ing tool as 1.3v. Set ATI Tool GPU volts to 1.3v will show OC'ing tool at 1.325v but the real voltage measured by DMM is closer to 1.3v as set in ATI Tool.

===================================
Benchmark results

Opteron 146 @ 10x200HTT = 2000mhz with Cat 5.11 drivers

vs

Opteron 170 @ 10x200HTT = 2000mhz with Cat 5.12 drivers

3dmark05
- Cat 5.12 on Opteron 146 resulted in 50pts less in 3dmark05 and 100pts less in 3dmark2003 at same clocks as Cat 5.11 drivers when tested.
- Opteron 146 tests used 2x 512MB Gskill PC4400LE TCCD 431 2-2-2-5 7-14-2222

@621/747 = 8,636 vs 8,911 (+275pts increase with Cat 5.12 /Dual core cpu)
@680(675)/800(792) = 8,986 vs 9,381 (+395pts increase with Cat 5.12 /Dual core cpu)
@710(708)/802(801) = 9,106 vs 9,572 (+466pts increase with Cat 5.12 /Dual core cpu)

3dmark2003
@621/747 = 16,368 vs 16,735 (+367pts increase with Cat 5.12 /Dual core cpu)
@708/801 = 17,698 vs 18,111 (+413pts increase with Cat 5.12 /Dual core cpu)

----------------------------------------------------------
Opteron 170 @10x260HTT = 2600mhz at 1.392v
2x 1GB OCZ PC4000EB Platinum @260mhz 3-3-2-5 7-14-2223

3dmark05
@621/747 = 9,284
@708/801 = 10,205
@756/900 = 10,965

3dmark2003
@621/747 = 17,847
@708/801 = 19,425
@749/900 = 20,535


----------------------------------------------------------
Opteron 170 @10x280HTT = 2800mhz at 1.552v
2x 1GB OCZ PC4000EB Platinum @254mhz 3-3-2-5 7-14-2223

3dmark05
@621/747 = 9,365
@708/801 = 10,310
@756/900 = 11,058


----------------------------------------------------------
Now compare that to these oc'd single core Opteron 146 cpu results
----------------------------------------------------------
Opteron 146 CAB2E 0540FPMW @10x280HTT = 2800mhz at stock vcore
2x 512MB Gskill PC4400LE @280mhz 2.5-3-3-6 7-14-2222

3dmark05
@621/747 = 9,235
@708/801 = 10,163
@742/891 = 10,795

3dmark2003
@621/747 = 17,707
@708/801 = 19,169


----------------------------------------------------------
Opteron 146 CAB2E 0540FPMW @10x300HTT = 3000mhz at 1.408V
2x 512MB Gskill PC4400LE @272mhz 2.5-3-3-6 7-14-2222

3dmark05
@621/747 = 9,287
@708/801 = 10,221
@749/882 = 10,869

3dmark2003
@621/747 = 17,945
@708/801 = 19,449
@735/882 = 20,328


----------------------------------------------------------
Opteron 146 CAB2E 0540FPMW @10x310HTT = 3100mhz at 1.536v
2x 512MB Gskill PC4400LE @282mhz 2.5-3-3-6 7-14-2222

3Dmark05
@621/747 = 9,321
@749/882 = 10,915

3dmark2003
@749/900 = 20,784


----------------------------------------------------------
Opteron 146 CAB2E 0540FPMW @10x313HTT = 3130mhz at 1.568v
2x 512MB Gskill PC4400LE @283mhz 2.5-3-3-5 7-14-2212

3dmark05
@756/900 = 11,049

3dmark2003
@756/900 = 20,860

----------------------------------------------------------

Well for 3dmark05 at least for max Opteron 146 /170 cpu clocks and @756/900 clocks for 512MB HIS X1800XT stock heatsink, the difference is ~9pts to the dual core Opteron 170

3dmark05 compare urls

Opteron 170 @2800mhz = 11,058

vs

Opteron 146 @3130mhz = 11,049




===================================

[deathman20]

Nice results soo far... Got any gaming benches like FEAR or something to show the difference between the 2? Be intresting to see if its really affected gaming speeds at all because I think it did with mine but can't be 100% sure switching from the 5.11's to 5.12's and im not switching back hehe.

[eva2000]

don't own FEAR, but probably see if i can whip up some gaming benchies after i get all the 3dmarks out of the way...

[dinos22]

yeap as i suspected........i think we'll be seeing some new records from you from 2.8GHz DC benchmarks

[Lightman]

that is very good sign to go for dual core
and this is olny beginning

[Unoid]

can't you feed ur 170 more volts? 1.50V and oc to 2.8ghz or so?

or is your prefab watercooling maxed out for temps?

[dinos22]

he can bench at 2.8GHz 1.55v

[perkam]

===================================
Overclocking X1800XT
I'm overclocking and changing voltages on X1800XT the same 5 step method i used on single core Opteron 146:

In service manage in windows disable ATI Smart and ATI Hotkey services and reboot pc. You loose temp reading in CCC but can use ATI Tool to view temps.
Use X1800 Oc'ing tool and set 3D desired clocks using the 2D panel
Use ATI Tool to only verify and view the 3D clocks are set correctly. So should be same as what you set in X1800 Oc'ing tool.
Whenever you set the mem clock, remember to set the GPU core clock again, or it will fall back to lower clocks and when you set the GPU core clocks again, remember to set the mem clock again, or the same will happen.
Setting voltages, OC'ing tool reports wrong voltages so DO NOT use OC'ing tool to set voltages, use ATI Tool to set voltages. Example, GPU voltage reported in ATI Tool is 1.275v will show in OC'ing tool as 1.3v. Set ATI Tool GPU volts to 1.3v will show OC'ing tool at 1.325v but the real voltage measured by DMM is closer to 1.3v as set in ATI Tool.

Nice Mini-Guide

Perkam

[eva2000]

can't you feed ur 170 more volts? 1.50V and oc to 2.8ghz or so?

or is your prefab watercooling maxed out for temps?

Yeah i can do 2800mhz at 1.55v benchable for Opteron 170 hence why i said more to come soon

[cantankerous]

Say Eva, what bios is that you are running on your DFI? 11/14-3? I have heard of no such bios and now feel really behind. I am still running 704-2BTA and thought that was the latest.

Where can I get the latest and do you know what changes the 11/14 bios contains over previous bioses? Thanks.

BTW, nice little thread. I can't wait to ditch my winnie and get an Opteron 175 in the new year.

[eva2000]

Updated thread with some 3dmark2003 results too.. more to come

canta, 1114-3 is on dfi official web site for NF4 UT SLI-DR product page it adds FX-60 dual core support i believe NF4 Ultra-D has 1114-1 or -2 listed some folks say it's same bios but to be safe i'd grab the 1114 one from the specific NF4 Ultra-D product page on dfi web site

[eva2000]

LOL updated with more results in first post, final max 3dmark05 was +9pts to dual core Opteron 170 hehe.. not much difference heh

[G H Z]

That last bench doesn't make alot of sense seeing your previous scores, also you we're running 1Gb on the 146

[eva2000]

not really look at opteron 170 @2.8ghz 621/747 vs opteron 146 @3.1ghz 621/747 was 9365 vs 9321

[NinjaWreck]

LOL updated with more results in first post, final max 3dmark05 was +9pts to dual core Opteron 170 hehe.. not much difference heh

Those results are using two different drivers, different amounts of ram, and different CPU speeds....

[perkam]

Click on edit, on the bottom right corner of your post ninja wreck, from there you can delete your dawble post

Let's see some 3.1ghz 170 action to really get this comparo interesting

Perkam

[ugp]

Hey eva2000 lend me your Opteron for a few days

[eva2000]

Those results are using two different drivers, different amounts of ram, and different CPU speeds....

that's the point with cat 5.11 on both dual core cpu looses out... with cat 5.12 drivers on both single core looses out... ram amount doesn't make much difference it's all about optimal clocks, drivers and settings for both rigs to see how they compare

you wouldn't compare X1800XT vs 7800GTX at the same clocks too would ya

Hey eva2000 lend me your Opteron for a few days

hehe...

[thelostrican]

that's the point with cat 5.11 on both dual core cpu looses out... with cat 5.12 drivers on both single core looses out... ram amount doesn't make much difference it's all about optimal clocks, drivers and settings for both rigs to see how they compare

you wouldn't compare X1800XT vs 7800GTX at the same clocks too would ya



hehe...

most definite but if your comparing the drivers you need to keep everything else the same...

[WiCKeD]

Good point as the average bencher will just run the latest CATS, but 50 points could be the difference from one run to the next and OCs will vary.

Excellent, thorough testing, eva!

[funkflix]

Any 01 benches? Dualcore should give u a nice boost, with the 5.12!

Great results!

[eva2000]

didn't do much 3d2k1 on single core rig so didn't have much data to compare against heh - only can try for max 3d2k1 score on dual core to see how it comes up against my current max for single core which is 40,094 http://service.futuremark.com/compare?2k1=8811044

[blind_ripper]

it will give u the same score as the singel i think on 2.9ghz , i went from 32.6k to 35.5k withe drivers , withe les gfx power

[funkflix]

I'm really intrested in a 01 run, with ur currently rig. I read something about a plus of around 4000 with dualcore and 01.

[eva2000]

well here's the best of a few runs i just did

146 @3130mhz ram at 283mhz 2.5-3-3-6 = 40,094
http://service.futuremark.com/compare?2k1=8811044

vs

170 @2800mhz ram at 254mhz 3-3-2-5 = 39,648
http://service.futuremark.com/compare?2k1=8822530

512 HIS X1800XT @756/900