• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

E7-2870

yea i would have expected around 26-28min TPF on those but its hard to tell what other factors may be at play here.

You are running linux right? the linux core was slower than windows when it first came it out so it very well could still be the case.
 
The memory speed can make a noticable difference. If this is running at DDR3-1066 with slow timings, that will make a difference. What are the memory timings and low-level memory benchmarks (aida/everest/etc) reporting?

It's highly unlikely you'll see chips in this range targetted for overclocking. Even the xeon 5500/5600 aren't marketed with this in mind. Large memory work-loads -- virtualization, database, etc, appear to be the primary target.
 
sfield said:
The memory speed can make a noticable difference. If this is running at DDR3-1066 with slow timings, that will make a difference.
Click to expand...
Precisely, I'm a bit surprised no one mentioned it before. -bigadv is very sensitive to memory performance.

It's highly unlikely you'll see chips in this range targetted for overclocking. Even the xeon 5500/5600 aren't marketed with this in mind. Large memory work-loads -- virtualization, database, etc, appear to be the primary target.
Click to expand...
Yes, even the dual-socket 6500-series Xeons which are essentially identical in most respects to the higher-end 7500-series, are niche marketed towards very high main memory environments even though they are limited to two sockets like their 5000 series siblings. That is the major difference between 6500 and the 5000-series.
 
APOLLO said:
Precisely, I'm a bit surprised no one mentioned it before. -bigadv is very sensitive to memory performance.
Click to expand...

Yea it matters a lot on Bigadv relative to regular SMP, but its not going to cause a 3-5min TPF discepancy.
 
APOLLO said:
Precisely, I'm a bit surprised no one mentioned it before. -bigadv is very sensitive to memory performance.
Click to expand...


i did mention it, i asked if it was running in quad channel DDR3, lol. my guess is its not running quad channel. quad channel alone would probably give it an extra 5-6k PPD. now if he is running quad channel DDR3 and hes getting those numbers, then the only other limitation would be the complete lack of L2 cache per thread.
 
I can see this being 10-15%, which is inline with 3 minutes or so. frequency, latency, and command rate all have additive small influences. ECC settings can also play a role. I saw this play out on my SR-2 as I moved to DDR3-1870 7-8-7-24-1T. Should check what the memory settings and command rate are and run some memory benchmarks.

Other memory access issues like interleaving configuration and NUMA local/remote memory access patterns (NUMA shouldn't apply for this case given 1P intel) can easilly add much more of a difference, particularly as more cores and NUMA nodes are added.

You may also want to play with options in the BIOS.

nitrobass24 said:
Yea it matters a lot on Bigadv relative to regular SMP, but its not going to cause a 3-5min TPF discepancy.
Click to expand...
 
8gb via 2x 4gb sticks so no not quad channel...
standard is 1333 cas 9 I am not sure if its using registered or ecc... but that slows it down too...

currently bigadv is limited to windows.... so this is win2k8r2 the other bigadv clients are on windows 7/8early alpha... I couldnt get the nic drivers to cooperate on this one so I grabbed 2k8r2...
I just got into the beta group... so I should be back on linux running the beta bigadv soon... I have always gotten rather nice numbers on linux...
 
Patriot said:
8gb via 2x 4gb sticks so no not quad channel...
standard is 1333 cas 9 I am not sure if its using registered or ecc... but that slows it down too...
Click to expand...


well then that explains the low number's. dual channel memory + low cpu clocks is killing the performance.
 
You must log in or register to reply here.
Back
Top