Navigate:

Advice
Beginners
BIOS Guide
CPUs
Links
Mainboards
Memory
Network
Storage
Video/Sound Cards

Contact
Forum
SiteMap
Sponsors
WebNews
Home
. .

Prices:

Mainboards

ABIT
ASUS
Chaintech
Shuttle
Soyo
Tyan

CPU
Intel
P4 2.4C-800
P4 2.6C-800
P4 2.8C-800
P4 3.0-800
P4 3.2-800

AMD
AthlonXP
XP 1700+
XP 2000+
XP 2400+
XP 2500+
XP 2700+
XP 3000+
XP 3200+

Athlon64
Athlon64 3200+
Athlon64 FX-51

Opteron
Opteron 240
Opteron 242
Opteron 244
Opteron 246

Memory

Corsair
Crucial
Kingston
Mushkin
OCZ

Search Prices:


























































































































LOSTCIRCUITS

SHORTCUTS:
Top page
The 64-bit Challenge
Registers, LOADs and STOREs
At One Glance
The Memory Controller
Latencies
Windows and WoW
Tunnels instead of Bridges
The Clocking Scheme
Crush K8 a.k.a. nForce3
The FX-51
Test Setups
Memory Benchmarks
Winstones
Caligari TrueSpace 5.1
Amorphium3
3DStudio Max 5.0
SPECapc's (3dsmax, Maya)
ViewPERF 7.0
Cinebench
3DMarks
Aquamark3, Comanche4, X2
UT2003, Gunmetal
MDK2, Flask 060
64-bit Performance
Encryption
Conclusion
Best Prices on Opterons and Athlon 64 Processors
Hit the Forums and let us know what you think
 AMD Athlon 64 FX-51   
New Frontiers
(Review by MS, September 23, 2003)
Memory Subsystem

We are using SiSoft Sandra's Memory Benchmark since it allows the use of all prefetch mechanisms that are necessary in order to get n idea about the raw bandwidth the system is capable of. The discussion has come up repeatedly that buffering should be disabled, however, we feel that those results are interesting but by no means relevant for the desktop platform.


Note the increase of memory bandwidth achieved by moving from 2.5:3:3 (FX51 2.5) to 2:3:2 (FX51 2.0)latency settings. At the PC3200 memory frequency, the system would not boot if tRCD was set to 2 cycles, even with Winbond BH-5 based DIMMs. The increase in bandwidth when the system was overclocked to a 210 MHz external clock signal shows that the memory bus at least is directly dependent on the CPU input clock. (D) dual channel nForce2, (S) single channel nForce2

All further benchmarks will be run at 2:3:2 unless specifically indicated otherwise.

Cachemem 2.65MMx

Lower is better; P4: transparent blocks, A64FX-51, petrol-color blocks. We only show the two top contenders here, that is the ASUS P4P800 with PAT enabled (whatever they may call it) which is the fastest P4 board out there still against the ASUS SK8N with the low latency memory controllers built into the CPU. Even though the P4 numbers are already out of this world, it does not stand a chance against the A64FX. Note the impact of the 1MB cache on the 1MB block size transfer latencies.

For the record, as much as we appreciate the existence of SiSoft Sandra and deem it a valid benchmark wherever appropriate, we are not going to show CPU benchmarks since they have no bearing on real world performance.

next page:    => Winstones =>

Click here! All advice and educational articles on LostCircuits are free, but if you feel you can, please make a small donation to us!
Thank you!

General disclaimer: This page only reflects the author's personal opinion and assumes no responsibility whatsoever regarding any of the contents or any damages that may occur explicitly or implicitly from reading the contents of this site. All names and trademarks mentioned in this review are the exclusive property of the respective parent companies.
All contents of this site are protected by international copyright laws. Reproduction of the contents even in parts is not allowed except after written permission by the author and referral to this site.
Copyright 2002 - 2008 LostCircuits