A Halloween Treat

Final benchmark effort with the FX51 @ 2604MHz ***1 hour 28 minutes**** Lets see what Hipro5 can do about that Here's some eye candy I snapped after the benchmark completed. I was pretty sure 2600 was possible with ice water, but just to see how far it would go I bumped it on up to 2700. I was able to run one benchmark and then it froze, so I'm satisfied that 2600 was about as much as I could get with my current cooling setup. Anyway here's what it does at 2700...

OK Now thats amazing! Great job!

wow! Way to go TC! Now for hipro to put an fx on ice...so to speak

TC, I wish congratulate you on the successful result of your efforts.

I don't remember, but did the extra cache on the P4 EE help SETI at all? I assume there were benchmarks done. If it does not help I don't think your score will have any competition for a while from that side of the fence at least. And if it does... you are basically just getting started here as far as extreme cooling. Seems like that chip has a lot of potential... much more than I would expected from such a new product.

I haven't seen any seti benchmarks from the P4 EE, but I'm sure the extra cache helps a fair amount. The Pentium M certainly benefits from it - my 1.6GHz PM laptop is only about 20 minutes shy of this FX bench while running 1000MHz slower.

TC said:

I haven't seen any seti benchmarks from the P4 EE, but I'm sure the extra cache helps a fair amount. The Pentium M certainly benefits from it - my 1.6GHz PM laptop is only about 20 minutes shy of this FX bench while running 1000MHz slower.

Click to expand...

Is the big cache the primary difference between the Pentium M and p4m or p3 chips? Just a theory here but I assume the reason it helps so much is precisely because of the low clock speed. Probably get a lot of benefit out of that cache that you might not see at 3.2 gigs or whatever.

No the pentium m is really more like a P3 with some of the bus enhancements from the P4 - like quad pumped bus, etc. It has a shorter pipeline and does more per clock while having some of the bandwidth features of a P4. Of course there's more to it than that, but that's the basic difference.

Here is a link for a p4ee vs. amd 64 & 64fx benchmarks- lots of pictures for benches on site and may take a few minutes for us dial up users to download. The seti bench is near the bottom of the page.

http://www15.brinkster.com/tbclb/AMD64P4EE/AMD64P4EE.htm

I'd post the seti bench pics to save some time but I'm not allowed to post attachments. Sorry. A brief run down of the listed performers according to the site-

4x opteron 1.8g-1800/2x166m---------0:34
4x xeon 2.8g-2800/100/2x100m-------0:39
2x xeon 3.06g2m-3066/133/2x133m--1:01
2x opteron 1.8g-1800/2x166m----------1:06
p4ee 3.2g-3200/200/2x200m------------1:17
2x xeon 3.06g-3066/133/2x133m-------1:20
(skipping a couple xeon entries...)
p4 3.2g 3200/200/2x200m----------------1:22
pr 3.00g 300/200/2x200m-----------------1:29
(skiping a couple of p4's...)
athlon fx-51 2200/2x200m-----------------1:45
athlon fx-51 2200/2x166m-----------------1:47
athlon fx-51 2200/2x133m-----------------1:48
p4 2.4g 2400/200/2x200m-----------------1:49
athlon fx-41 2200/2x100m-----------------1:51
athlon 64 3200+-2000/1x200m-----------1:58
athlon xp 3200+-2200/200/2x200--------2:11
(skipping a few more athlon/p4 listings...)
athlon xp 2400+-2000/133/2x133m------2:42
p4 2.53g-2533/133/2x166m----------------2:44
pr 2.40g-2400/133/2x166m-----------------2:48

TIME (h:mm)

In a small box it lists seti info and that reads-

client: v 3.03 (text version)
work unit with angle range: 0,417
HT: bech with two same work units

Under the box w/seti info is some other info and I'll quote that here-

There's no beating the P4's HyperThreading: even the 64-bit version delivered no better results with Linux - using the Athlon 64 FX-51 and Athlon 64 3200+. "The New Intel P4 EE Take over the AMD 64 FX"

Click to expand...

Hope that helps TC along with everyone else, and I am sorry for typing a large message.

-Tyrinon

That's Tom's info - he doesn't bench like normal folks so it's hard to draw any comparisons. We need to see the time for a single client, not 2 run at the same time divided by some figure of his choosing.

lol. oops. Sorry 'bout that. I'll see what I can dig up...

-Tyrinon

Great work TC!

Cy

WOW!!! Nice bench!!