elfuego wrote:

Interesting fact is that replacing those Xeons with Tualatins 1.4 would literally triple the result. 😄

Yeah... but then it wouldn't be a Xeon rig 🤣.

I'm trying to find a pair of 900/2MB Xeon to swap in.

I don't get this hype to upgrade or overclock a good setup retro PC even more. Usually the system is already good as it is from the retro point of view. If I want a faster system I just pick another CPU from the pile (last estimation 150+) or simply sit in front of my main system. For me it is much more interesting to see comparisons of some stock systems.

elianda wrote:

I don't get this hype to upgrade or overclock a good setup retro PC even more. Usually the system is already good as it is from the retro point of view. If I want a faster system I just pick another CPU from the pile (last estimation 150+) or simply sit in front of my main system. For me it is much more interesting to see comparisons of some stock systems.

I like to max out my retro rigs, but I do like to stay true to the original platform I started a build with. So for this rig, it's the 440GX chipset.

m1919 wrote:

Yeah... but then it wouldn't be a Xeon rig 🤣.

I'm trying to find a pair of 900/2MB Xeon to swap in.

Interesting rig. Are the 1MB/2MB caches on the Cascades processors on-chip or is it just full speed cartridge-mounted cache? 2MB would've been a monstrous amount of on-chip memory in 2000!

Standard Def Steve wrote:

m1919 wrote:

Yeah... but then it wouldn't be a Xeon rig 🤣.

I'm trying to find a pair of 900/2MB Xeon to swap in.

Interesting rig. Are the 1MB/2MB caches on the Cascades processors on-chip or is it just full speed cartridge-mounted cache? 2MB would've been a monstrous amount of on-chip memory in 2000!

The P3 Xeons up to 550Mhz have full speed off-die cache, P3 Xeons 600Mhz+ have on-die cache of 1-2MB on the 100Mhz FSB versions and 256Kb on the 133Mhz FSB Versions.

Heh, gonna take a cue from luckybob. The 256Kb versions are "Pseudo Xeons" as they're pretty much just regular Coppermines in a Slot-2 package.

Xeon Prime: ASUS XG-DLS, Dual P3 Xeon 700/1MB SL49P, 1.5GB PC100 ECC, GeForce4 Ti4600 128MB, Windows 2000 Pro

With the same 4600 Ti and a 1.4S Tualatin @ 1.6ghz, I got 9715 marks in 3d mark 2001SE.

sgt76 wrote:

With the same 4600 Ti and a 1.4S Tualatin @ 1.6ghz, I got 9715 marks in 3d mark 2001SE. […]
Show full quote

With the same 4600 Ti and a 1.4S Tualatin @ 1.6ghz, I got 9715 marks in 3d mark 2001SE.

I wonder how a pair of 900/2MB Xeons would do.

Also, would there be any way to force 3dMark to run on my Obsidian2? I remember seeing someone benchmark a pair of Voodoo2s using it, but I have no options available to let me choose my card.

By any chance, were they mine perhaps? I used powerstrip to alternate between the AGP card and my V2 SLIs.

3dmark 2000 - 2856 marks (1024x768, 16 bit, default settings)

3dmark 2001se - 1131 marks (1024x768, 16 bit, default settings)

All benches done with Voodoo 2 SLI at an overclock of 95mhz

sgt76 wrote:

With the same 4600 Ti and a 1.4S Tualatin @ 1.6ghz, I got 9715 marks in 3d mark 2001SE.

Hmm... I must admit that I'm a little disappointed. I expected more from Tualatin. Ah well... I'll stick to Socket A 😀 Thanks for the benchmark!

Edit: BTW, I suppose you did the test with SDRAM (i815)... Can anyone provide a test with dual tualatins or tualatin with DDR or rambus?

I was using sdram @ 152mhz. ddr apparently won't make much difference as the P3's bus is too slow for ddr bandwidth to be effectively utilized.

Cpu benchmarks is where the Tualatin really shines, performing like a NW 'B' @ 2.8ghz. Anything FPU related, not that hot... maybe like a 1.8ghz Willamette. 😜

elfuego wrote:

sgt76 wrote:

With the same 4600 Ti and a 1.4S Tualatin @ 1.6ghz, I got 9715 marks in 3d mark 2001SE.

Hmm... I must admit that I'm a little disappointed. I expected more from Tualatin. Ah well... I'll stick to Socket A 😀 Thanks for the benchmark!

Edit: BTW, I suppose you did the test with SDRAM (i815)... Can anyone provide a test with dual tualatins or tualatin with DDR or rambus?

I don't know if anyone saw my P3-S (i815) result on the first page, but I got 11,312 with a 9800Pro. A Willamette 1.8 (DDR266) with the same video card only scored 9147.

sgt76 wrote:

I was using sdram @ 152mhz. ddr apparently won't make much difference as the P3's bus is too slow for ddr bandwidth to be effectively utilized.

I heard about this rumor, but I find it still hard to believe. I remember the difference changing from KT133a to KT266a on athlon XP 1600+, the frame rate in Morrowind seemed to double. Thats why I hoped it would be similar on tualatin...

Abysmal Windows 7 i7 920 w/ GTX 690. I didn't see one up so I thought I'd try, terrible results.

I should note that I receive this message at startup:

I get the same message with SLI disabled and I score 37094 that way.

On my C2D 7300 w XP and a GTX580 I get : 47757

Here my i7 3770 + GTX660:

I think 2001 is still the best 3DMark in terms of interesting graphics. It just never gets old. All the new ones are so boring to watch.

Shagittarius wrote:

Abysmal Windows 7 i7 920 w/ GTX 690.

I dont understand, do you mean abysmal windows or abysmal i7 920? Or is it GTX690 that's abysmal? Where does the problem lie? 😕

I get the same message at startup!

There's nothing wrong with any of the components. Sure the i7 920 is a little underpowered for the 690 but something is wrong with drivers or the benchmark and that particular card it seems. It performs like I would expect in real world applications.

Like I said I tried it un-SLI'ed and it only scored a little more than with which is to be expected.

Win 7 just isn't very good at supporting directx 8 stuff. Drivers are not optimized for it anymore either. I'm pretty sure in XP I got more like 60k vs the 40k I get in win7.

Interesting enough in 3dm2003 I get about 80k in windows 7 =]

I can't even run the damn thing on my Win 7 64 bit setups. How do you get this to run on Win 7 64 bit? 😢