VOGONS


Battle of the platforms: socket 939!

Topic actions

Reply 560 of 577, by nd22

User metadata
Rank l33t
Rank
l33t

The 4 processors ready to go:
A. Athlon 64 X2 3800;
B: Athlon 64 X2 4200 - forgot to take to screenshot with the Everest and CPU-Z, I will take it tonight, for now a screenshot with the system details and PCMARK 2004;
C. Athlon 64 X22 4400 - there are 2 revisions: ADA4400DAA6CD has a 110W TDP and it is the one I got; the second one is ADV4400DAA6CD with a 89W TDP. There is not much difference between them, all Athlon 64 X2 are very easy to cool with either Arctic freezere 64 pro or the AMD stock cooler supplied with AM2 processors!
D. Athlon 64 X2 4800 - top of the line from the regular line of CPU's.

Reply 561 of 577, by nd22

User metadata
Rank l33t
Rank
l33t

Some of you might think: where is the Athlon 64 X2 3600, the smallest model?
Disabling the cache to only 2*256kb has a detrimental effect on performance; you are getting a dual core with a Sempron amount of L2 cache. AMD should just called it Sempron X2 and not Athlon 64 X2! If you are building a high end socket 939 system get yourself a real X2 not a surrogate!

Reply 562 of 577, by nd22

User metadata
Rank l33t
Rank
l33t

The benchmarks stay the same; the order in which the screenshot are put up will always be:
A. Athlon 64 X2 3800
B. Athlon 64 X2 4200
C. Athlon 64 X2 4400
D. Athlon 64 X2 4800

Reply 563 of 577, by nd22

User metadata
Rank l33t
Rank
l33t

The first test is Aquamark3, the first DirectX9 benchmark ever released! In all resolutions and settings we are CPU-limited which is a good thing as this is exactly what we want to see: how the dual cores line up!
1. default score first - the average FPS is what we are after
A. Athlon 64 X2 3800
B. Athlon 64 X2 4200

Reply 564 of 577, by nd22

User metadata
Rank l33t
Rank
l33t

1. C. Athlon 64 X2 4400
D. Athlon 64 X2 4800

Reply 565 of 577, by nd22

User metadata
Rank l33t
Rank
l33t

At default settings the difference between the weakest and the strongest model is only 14%, far below the 26% the PR rating implies!

2. At the same 1024*768 resolution but at max details and with sound we got:
A. Athlon 64 X2 3800
B. Athlon 64 X2 4200

Reply 566 of 577, by nd22

User metadata
Rank l33t
Rank
l33t

C. Athlon 64 X2 4400
D. Athlon 64 X2 4800

Reply 567 of 577, by nd22

User metadata
Rank l33t
Rank
l33t

Again the delta between the 3800 and the 4800 is far below the difference between the PR ratings of these processors!

3. 1280*1024 is next, without sound first:
A. Athlon 64 X2 3800
B. Athlon 64 X2 4200

Reply 568 of 577, by nd22

User metadata
Rank l33t
Rank
l33t

C. Athlon 64 X2 4400
D. Athlon 64 X2 4800

Reply 569 of 577, by nd22

User metadata
Rank l33t
Rank
l33t

4. ... and with sound:
A. Athlon 64 X2 3800
B. Athlon 64 X2 4200

Reply 570 of 577, by nd22

User metadata
Rank l33t
Rank
l33t

C. Athlon 64 X2 4400
D. Athlon 64 X2 4800

Reply 571 of 577, by nd22

User metadata
Rank l33t
Rank
l33t

5. The final resolution to be tested is 1600*1200!
no sound:
A. Athlon 64 X2 3800
B. Athlon 64 X2 4200

Reply 572 of 577, by nd22

User metadata
Rank l33t
Rank
l33t

C. Athlon 64 X2 4400
D. Athlon 64 X2 4800

Reply 573 of 577, by nd22

User metadata
Rank l33t
Rank
l33t

6. And the most demanding settings: 1600*1200 max details, sound enabled:
A. Athlon 64 X2 3800
B. Athlon 64 X2 4200

Reply 574 of 577, by nd22

User metadata
Rank l33t
Rank
l33t

C. Athlon 64 X2 4400
D. Athlon 64 X2 4800

Reply 575 of 577, by nd22

User metadata
Rank l33t
Rank
l33t

There is not a single instance where the 4 models of Athlon 64 X2 tested scale according to the PR rating of these processors! The maximum difference between the 4800 and 3800 is 18.59% at the very last settings yet the PR difference is 26%!
So aquamark3 does not scale according to the AMD own PR ratings!
Let's continue with 3dmark 2001 - a benchmark that is mandatory to have in all retro reviews!

Reply 576 of 577, by nd22

User metadata
Rank l33t
Rank
l33t

As promised, the screenshot with the Athlon 64 X2 4200 - another Toledo based processor; a downside of all 4 configurations is the 2T command rate that has a large effect on performance! Registered RAM is required in order to have 4 modules run with command rate 1T! With an X2 is very easy to run 4 1GB memory modules at DDR-400 with no problems whatsoever:

Reply 577 of 577, by AlexZ

User metadata
Rank Oldbie
Rank
Oldbie

Having 4 DDR400 ECC registered modules is unnecessary in s939 as those are available in 2-4GB capacity per stick. 2 GB ones do not have heatsinks and do not appear to need them. I have one 4GB stick but never tested it as I don't want to risk my cpu/board with it.

Those interested to go above your benchmark results and run Command Rate 1T with ECC registered modules should get 4x more sticks than they need for DDR480 speed and 8x more sticks than they need for DDR533 speed for binning purposes. At DDR533 speed AIDA64 memory latency should be about 54ns and memory bandwidth about 8300MB/s. With only 2 sticks very high frequency like DDR480 should be possible.

Pentium III 900E,ECS P6BXT-A+,384MB,GeForce FX 5600, Voodoo 2,Yamaha SM718
Turion 64 MT-40@2.4Ghz,Gigabyte GA-K8NE,2GB,GeForce GTX 275,Audigy 2ZS
Phenom II X4 965,Gigabyte GA-MA770-UD3,8GB,GeForce GTX 780
Vishera FX-8370,Asus 990FX,16GB,GeForce GTX 980 Ti