Athlon 64

[Arminiusc]

here is a link to a benchmark of the Athlon 64 on a message board
http://www.tbreak.com/forums/showthread.php?s=&threadid=19317

[Arminiusc]

Found another link that is going to have information on the Athlon FX later today or tommorrow.
http://www.hardwareanalysis.com
They have a short article right now but they are working on a bigger review that people might be interested in.

[Madcat]

Should prove quite interesting. :D

[numibesi]

WOW !!!

I really would love to have one of those, did i saw right?!?!?!

A64Bit at 2.4Ghz
3DMARK2001 -> 24825 pts

P4 at 4,1Ghz
3DMARK2001 -> 24747

"ALWAY'S BOMBATING....."

[Bionic_Redneck]

3dmark is a bogus benchmark for real cpu power it's all based on the video card and memory bandwidth. prime or the dnet clients would be much better for that.

[Arminiusc]

3dmark isnt totally bogus... it is a good starting point for comparing... but you do need more specific tests. I cant wait to see more benchmarks on the 64 to see if I need to think about upgrading. :)

[Bionic_Redneck]

Prime is the client for grimps and typically P4 are the fastest because of SSE2. I would like to see how well the cpu does. I can't really think of any DC client that really take avantage of SSE2 other than that .

3dmark has it's use but it iwas meant for benchmarking video cards maybe passmark would be more useful to decide if you wanted to upgrade. Doesn't sissoft sandra have a benchmarking section in it?

[Arminiusc]

I think sysoft only benches floating point and integer... not sure which method they use...

[Arminiusc]

http://www20.tomshardware.com/cpu/20030217/cpu_charts-23.html
This is a link to a page of an older article at tomshardware that shows 3dmark does help determine cpu power... even though you have a fancy video card, the cpu still needs to be able to keep up. They used a radeon 9700 pro and 512mb ram in all the systems they compared.
I found it interesting to see how all the new cpus+m/b's compared to the old cpus+m/b's.

[Bionic_Redneck]

hey if you take a close look at that graph it proves my point. 3dmark is all about the video card and memory bandwidth. They have a athlon 1000 w/ 133 sdram out doing a athlon 1200 /w 100 sdram. also thats why athlon64 does so well even though it's has some awesome power it also is no slacker in the memory bardwidth area eithier.

[Arminiusc]

Yes you definately have a valid point. 3dmark benchmarks the whole system with some preferences to the video card. The only time the cpu doesnt really matter in 3dmark is when the video card is maxed out and a faster cpu doesnt do any good. The latest video cards though are cpu limited. We already know that the athlon performs better with the higher fsb also.
Of course for DC projects sisofts drystones and whetstones would give a better indication of performance.
As always though... the best way to see if something does what you want is to try it yourself! :)
Is anyone going to buy one when its avaiable?

[Bionic_Redneck]

I don't plan on buying an opteron but will be needing another cpu as I still am planning on building a small diskless node cluster. Been thinking about getting one of the new duron 1800's if they are reasonable priced.

[Arminiusc]

I read somwhere that the durons are hackable to activate the full cache instead of the half-cache they come with, but some durons dont work well due to the fact that the other half of cache is faulty on some.

[Madcat]

It's funny you guys mention benchmarks and accuracy. The only one I know that isn't influenced by manufacturers (or so I read) is the spec.org benches like specviewperf. They do have a seperate cpu bench in thier lineup or the COSBI benches.

Sandra is a good general bench but tends to be more of a quick and dirty kind of test that doesn't put any one component through the wringer thoroughly, since it uses only a tiny bit of code that loops it over and over.

Found another one that may be of interest. I thought this one had died off. I guess they were just waiting for the AMD64 capable cpu's and compilers to come out. :D

[Bionic_Redneck]

I read somwhere that the durons are hackable to activate the full cache instead of the half-cache they come with, but some durons dont work well due to the fact that the other half of cache is faulty on some.

thats new to me but I have heard that the you can re-enable sse in duron morgans with wcpuid so it might also work for the applebred's too.