X2 3800 @ 2.6 Benchmarks

D

DefJae

n00b
Joined
Mar 12, 2006
Messages
52
Ok guys I finally reach the end of my testing and research. Alot of hours was put into this so I hope you guys enjoy it. Let this be a guide or reference to those who are curious as to what these chips can do. By far this little adventure was nothing short of amazing and a whole lot of WOW :eek: I'm pretty sure if I put a little more effort I might be able to hit 3.0, but we'll save that for the next adventure and when I get a better sticks. I will post screen shots upon request.

OC @ 2.6

Sandra Sisoft

CPU Arithmetic:
Dhrystone ALU - 21998
Whestone FPU/iSSE2 - 8242/10736

Memory Bandwidth:
Ram Bandwidth Int Buff'd iSSE2 - 6224
Ram Bandwidth Float Buff'd iSSE - 6184

EVEREST

Memory Read - 6592
Memory Write - 6735
Memory Copy - 5672
Memory Latency - 47.8

CPU Queen - 5003
CPU Photoworxx - 20622
CPU ZLib - 30429
FPU Julia - 2935
FPU Mandel - 4343
FPU SinJulia - 2704

OC 2.8ghz

Sandra Sisoft

CPU Arithmetic:
Dhrystone ALU - 23686
Whestone FPU/iSSE2 - 8879/11564

Memory Bandwidth:
Ram Bandwidth Int Buff'd iSSE2 - 5939
Ram Bandwidth Float Buff'd iSSE - 5944

EVEREST

Memory Read - 6131
Memory Write - 6227
Memory Copy - 5789
Memory Latency - 50.1

CPU Queen - 5386
CPU Photoworxx - 19844
CPU ZLib - 32967
FPU Julia - 3157
FPU Mandel - 4675
FPU SinJulia - 2912
----------------------------
BFG 7800GT OC @ Core 470 - Mem 1120

3DMark03 @ 2.6 - 15364
3DMark05 @ 2.6 - 7001
-----------------------------
BFG 7800GT OC @ Core 470 - Mem 1120

3DMark03 @ 2.8 - 17112
3DMark03 @ 2.8 - 8038

BIOS Settings @ 2.6

CPU Speed:
CPU Clock 2600.3 MHz (original: 2000 MHz, overclock: 30%)
CPU Multiplier 10.0x
CPU FSB 260.0 MHz (original: 200 MHz, overclock: 30%)
Memory Bus 216.7 MHz
DRAM:FSB Ratio CPU/12

Memory Timings:
CAS Latency (CL) 2T
RAS To CAS Delay (tRCD) 3T
RAS Precharge (tRP) 3T
RAS Active Time (tRAS) 6T
Row Cycle Time (tRC) 8T
Row Refresh Cycle Time (tRFC) 16T
Command Rate (CR) 1T
RAS To RAS Delay (tRRD) 2T
Write Recovery Time (tWR) 2T
Read To Write Delay (tRTW) 3T
Write To Read Delay (tWTR) 2T
Write CAS Latency (tWCL) 1T
Refresh Period (tREF) 200 MHz 15.6 us
DQS Skew Control Disabled
DRAM Drive Strength Weak
DRAM Data Drive Strength 1 (50% Reduction)
Max Async Latency 7 ns
Read Preamble Time 6.5 ns
Idle Cycle Limit 256
Dynamic Idle Cycle Counter Disabled
Read/Write Queue Bypass 16
Bypass Max 7
32-byte Granularity Disabled

Sensor Properties:
Sensor Type ITE IT8712F (ISA 290h)
GPU Sensor Type Diode (NV-Diode)
Motherboard Name DFI LANParty UT nF4 SLI-DR Expert
Chassis Intrusion Detected Yes

Temperatures:
CPU 29 °C (84 °F) - Load is about 38c
Chipset 50 °C (122 °F)
PWM 45 °C (113 °F)
GPU Diode 42 °C (108 °F)
Seagate ST380011A 34 °C (93 °F)
WDC WD400BB-00JKC0 41 °C (106 °F)
WDC WD3000JD-00KLB0 34 °C (93 °F)

Cooling Fans:
System 1974 RPM
Chipset 4327 RPM

Voltage Values:
CPU Core 1.33 V
+3.3 V 3.26 V
+5 V 4.92 V
+12 V 12.16 V
+5 V Standby 4.92 V

BIOS Settings @ 2.8

CPU Properties:
CPU Type DualCore AMD Athlon 64 X2
CPU Alias Manchester S939
CPU Stepping BH-E4
CPUID CPU Name AMD Athlon(tm) 64 X2 Dual Core Processor 3800+
CPUID Revision 00020FB1h

CPU Speed:
CPU Clock 2800.8 MHz (original: 2000 MHz, overclock: 40%)
CPU Multiplier 10.0x
CPU FSB 280.1 MHz (original: 200 MHz, overclock: 40%)
Memory Bus 200.1 MHz
DRAM:FSB Ratio CPU/14

Temperatures:
CPU 30°C (84 °F) - Load is about 38c
Chipset 47 °C (117 °F)
PWM 46 °C (115 °F)
GPU Diode 43 °C (109 °F)
Seagate ST380011A 32 °C (90 °F)
WDC WD400BB-00JKC0 39 °C (102 °F)
WDC WD3000JD-00KLB0 33 °C (91 °F)

Cooling Fans:
System 2096 RPM
Chipset 3924 RPM

Voltage Values:
CPU Core 1.39 V
+3.3 V 3.26 V
+5 V 4.92 V
+12 V 12.16 V
+5 V Standby 4.92 V
VBAT Battery 3.10 V

Memory Timings:
CAS Latency (CL) 2.5T
RAS To CAS Delay (tRCD) 3T
RAS Precharge (tRP) 3T
RAS Active Time (tRAS) 7T
Row Cycle Time (tRC) 8T
Row Refresh Cycle Time (tRFC) 16T
Command Rate (CR) 1T
RAS To RAS Delay (tRRD) 2T
Write Recovery Time (tWR) 2T
Read To Write Delay (tRTW) 3T
Write To Read Delay (tWTR) 2T
Write CAS Latency (tWCL) 1T
Refresh Period (tREF) 200 MHz 15.6 us
DQS Skew Control Disabled DRAM Drive Strength Weak
DRAM Data Drive Strength 1 (50% Reduction)
Max Async Latency 7 ns
Read Preamble Time 8.0 ns
Idle Cycle Limit 256
Dynamic Idle Cycle Counter Disabled
Read/Write Queue Bypass 16
Bypass Max 7
32-byte Granularity Disabled

Disclaimer
Let this be a guide and reference, but by any means I guarantee these numbers and settings will reflect or work on a similar hardware. Final OC are tested with OCCT 30 min. test, Memtest Test No.5 40 pass No error, and 10hrs x2 prime95 In place large fft. All benchmarks are performed in my best abilities and knowledge.

This would not be possible with out these GREAT Guides:
http://www.hardforum.com/showthread.php?t=888001
http://www.dfi-street.com/forum/showthread.php?t=20823
 
jcll2002 said:
youre also water cooled :(
Click to expand...


That really shouldn't affect his OC until he starts upping the core voltage. If it will do 2.6ghz at stock voltage then it probably would do the same on good regular cooling like a SI-120. Nice going!!


BTW - unless there has been a change in the rules that I haven't noticed, the OP needs to change his sig down to 10 lines or less.
 
jcll2002 said:
woah! you could prob go higher with only that little amt of voltage
Click to expand...

2.8 is def. do able, only problem I had is that my mem can't keep up. I had to lower my mem divider down to 150. I'm planning on getting a better memory, this should fix some problems.
 
Yup my dell x2 3800 is doing 2750 dual prime95 stable. This is on stock voltage air cooled. Haven't really messed with changing the voltage yet.
 
DefJae said:
2.8 is def. do able, only problem I had is that my mem can't keep up. I had to lower my mem divider down to 150. I'm planning on getting a better memory, this should fix some problems.
Click to expand...


I wouldnt say definately, ive had several cpu's in the past that would clock quite high at low voltage but would hit a wall, and regardless of the voltage bump, just simply wouldnt get higher. Im sure its more than likely that 2.8 is do'able, but thought id share my knowledge :)
 
wee96 said:
I wouldnt say definately, ive had several cpu's in the past that would clock quite high at low voltage but would hit a wall, and regardless of the voltage bump, just simply wouldnt get higher. Im sure its more than likely that 2.8 is do'able, but thought id share my knowledge :)
Click to expand...

I was able to hit 2.8 with stock vcore and I able to do some benchmarks. Not to say it is stable in stock vcore, it is still pretty impressive by any means. Cheers.....
 
jcll2002 said:
but do you think im likely to get 2.6+? its on the dfi ultra nf4 infinity btw. Doesnt it just depend on the stepping?
Click to expand...

What stepping do you have? That mobo should be good plus 2.6 is the average for 3800 X2's. Try to read up there is alot of good read in this forum and in dfi-street
 
Here are some screen shots of 3dmark03 and 3dmark05. I think I should have gotten a better score in both of'em, but I guess it will do for now.



 
Its a good thing these A64 chips clock as well as they do, because they are unbearably slow in the 1.8 - 2.2ghz range :) There's no way I could use my X2 3800 at 2.0 ghz. Even with the extra core, it still felt considerably 'slower' than my ancient (single core obviously) XP2500 at 2.3 ghz. Now my X2 at 2.8ghz? Mmmm...nice.
 
homersapien said:
Its a good thing these A64 chips clock as well as they do, because they are unbearably slow in the 1.8 - 2.2ghz range :) There's no way I could use my X2 3800 at 2.0 ghz. Even with the extra core, it still felt considerably 'slower' than my ancient (single core obviously) XP2500 at 2.3 ghz. Now my X2 at 2.8ghz? Mmmm...nice.
Click to expand...

These chips are pretty darn good. I will work on getting my X2 to be stable at 2.8. Only thing is as I stated previously I am debating on lower mem divider and looser timing oppose to tighter timings and high dividers. With higher OC i know difference is hardly noticeable. Ehhhh what to do what to do.
 
Ah ha! Here you are Defjae. I saw your post on dfi-street. Was following the thread there...

re: your screenshots of 2.8 I see that cpu-z reads the vcore at 1.36. I know you are using a dfi board and that it tends to undervolt. What vcore settings did you use for your dfi bios to get 2.8?

My chip is almost exactly the same performer as yours up to 2.6
 
Medstu said:
Ah ha! Here you are Defjae. I saw your post on dfi-street. Was following the thread there...

re: your screenshots of 2.8 I see that cpu-z reads the vcore at 1.36. I know you are using a dfi board and that it tends to undervolt. What vcore settings did you use for your dfi bios to get 2.8?

My chip is almost exactly the same performer as yours up to 2.6
Click to expand...

I actually left it at Auto and 102%. I can ran prime up till 5 hours but i'll end up with 1 error, pass the OCCT 30min test, and ran mem test #5 for 40 pass no errors. I was able to stablize it if I up the vcore a tad. I only did it to compare performance difference between 2.6 higher mem divider and tighter timings or 2.8 lower dividers and looser timings. The result is pretty impressive, only thing is running your mem at lower dividers and looser timings ends up being your bottle neck. I rather keep it at 2.6 higher divider and tighter timings, plus CPU performance difference between the two is not much. I will do some more testing after work.

I had a mishap this morning, I accidently unplug the power for my pump, luckly CPU got to hot and shut of by itself :( . Not sure if that will affect my OC or not, but I have a feeling it is not a good thing. I'll keep you guys posted and hopefully i can post some more results for you guys.
 
DefJae said:
These chips are pretty darn good. I will work on getting my X2 to be stable at 2.8. Only thing is as I stated previously I am debating on lower mem divider and looser timing oppose to tighter timings and high dividers. With higher OC i know difference is hardly noticeable. Ehhhh what to do what to do.
Click to expand...

Probably won't make any difference, unless your mobo has a problem running certain dividers; rare, but it happens.
 
Whew! ok I finally finish all my testings and benchmarks. I updated my original post in page 1 with all the goodies. I hope you guys enjoy it as much as I have. Here are my final benchmarks for 3dmark03 and 3dmark05. Cheers...

Final OC @ 2.8 3Dmark03



Final OC @ 2.8 3Dmark05

 
jcll2002 said:
so getting ram with tight timings really does matter?
Click to expand...

IMO a ram that can be run at 1:1 with your current CPU Speed is the way to go. Your ram will always be the bottleneck if you run it at a very low dividers and timings. Slower ram means that your CPU will have to wait for its to process data before execution. Some thing like that.
 
so would it be better to get ram w/ tighter timings or a pc4000 stick? ocing of course ;)
 
jcll2002 said:
so would it be better to get ram w/ tighter timings or a pc4000 stick? ocing of course ;)
Click to expand...

PC4000 is good, gives you more head room for OC, although it's not tight, but you will be able to run it close to 1:1. I rather have it loose with a higher divider or 1:1. This will let your memory run along side your CPU speed. There is alot of discussion about tight, loose, high dividers and 1:1 in this forum.
 
cpuz7ch.jpg

dang u hit 2.8

i'm really tempted to try it :eek:
 
Which G-Skill do you have? Mine seems to max out near 240mhz or so. I don't remember what the model number is, but it came with red heat-spreaders.
 
What kind of minimum OC can one expect from a same-core x2 3800+? Would getting the wrong stepping really screw me?
 
You must log in or register to reply here.
Back
Top