• A Great friend to the HardForum with a great kid that he is trying to get a scholorship to continue his schooling. Please give hime a vote! Only 24 hours left! Thanks.
    If you have an VOTE FOR KEENAN!

CPU OC with SMP/Bigadv TPF & PPD Database

I just uploaded the benchmarks that HFM has collected on my SMP clients over the last 3-4 days. I didn't see your request not to put the full suite up there lol. Feel free to delete entries on WU's that aren't of interest.
 
Yeah, more data. I want to be able to go in there and see the effects of lynnfield ram timings etc. on 6900 TPFs.
 
I will add memory timings and speed to my entries tomorrow.

Interesting comparing the 3ghz Q6600 in my HTPC to nightbird's entry on the 6701 work unit, it looks like the GT240 costs roughly 500 ppd from the SMP to run. It generates about 4200 pdd, so my net gain is 3700...not bad for about 40 watts.
 
I think it is of interest to have them all - it is not up to me - just delete the word "bench" from the unit column - unless you are using Musky's benchmark, but looking at the spread of units you aren't.

The "bench" is meant so that people know it is the same captured work unit without the variation you see in units 'in the wild'.

(Benchmark available on request ;))
 
i would get rid of the duplicates from the same family go with the higher PPD one. so the 6701 on the Q660 i'd take that off and use the 6702. the 6066 on the x4 630 i'd take that off and use the 6067 thats there instead.
 
sirmonkey1985 said:
i would get rid of the duplicates from the same family go with the higher PPD one. so the 6701 on the Q660 i'd take that off and use the 6702. the 6066 on the x4 630 i'd take that off and use the 6067 thats there instead.
Click to expand...

Wouldn't it be more valuable to have both entries for the family? That way if we have enough data, you could find the average PPD for the family?

I realize that the data here is all inherently 'ballpark' because there are so many variables but I'm always one for having as much valid data as possible and then narrowing it down through filters and quantitative analysis when you want your specific points.

If you would really like them removed though I will take them off.
 
well the reason i say that is because for example in your numbers they very a bit which means theres to many variables in it. for example there is almost a 20 second difference in TPF between the 6066 and 6067 which could pertain more toward the WU the gtx 460 is running. they should generally be within a couple seconds of each other. so it makes 1 WU look better then the other when thats not really the case. but either way you can keep them there it at least shows how much of an effect the GPU client has on the SMP client.

since you added the 2633 WU i decided to add mine as well since they are rare WU's that are quad core only. just wish there were more of them. love those WU's.
 
Yeah I like those too, for some reason only my Athlon seems to get them and my C2Q systems have been pumping 67xx units.

You make a good point with the work unit families and I will average them to bring some more stability to my numbers. I will also make a point to come back and update with new averages after I have more than 2 weeks on my clients.

Maybe a column for number of frames observed would help gauge data validity as well?
 
R-Type said:
Yeah I like those too, for some reason only my Athlon seems to get them and my C2Q systems have been pumping 67xx units.

You make a good point with the work unit families and I will average them to bring some more stability to my numbers. I will also make a point to come back and update with new averages after I have more than 2 weeks on my clients.

Maybe a column for number of frames observed would help gauge data validity as well?
Click to expand...


the number of frames observed is always 300 in HFM no matter how many of them you run. the GPU's vary depending on the size of the WU.
 
jebo_4jc banned?:eek:

What happened? I am in shock...

5 years and gone, just like that? :(
 
oh ok. :(

You know, I seem to miss all the bad stuff that goes on around here - I wander around in my [H]appy place completely oblivious to any fighting, which means either the mods here are awesome or I am blind. Or both.

But the reason I spend so much time here is it is by far the most pleasant and constructive place I have found on the interwebs, ever.

Righto then... thanks for the update.
 
I think it was in the graphics forum or some other forum like it that is constantly filled with too much troll testosterone. :rolleyes:
 
almost all of the forum is filled with stupid trolling kids except the DC section and genmay.
 
That shocks me, jebo is one of the most pleasant guys to deal with on the boards.

I have seen plenty of threads (particuarly dual vs quad and ati vs nvidia) degrade into shouting matches and name calling without any bans, I can't imagine what happened.

It will be good to have him back though.
 
Added and updated some more information from my tweaked SR-2 Rig, numbers are looking good! :D
 
Chilly said:
Added and updated some more information from my tweaked SR-2 Rig, numbers are looking good! :D
Click to expand...

Looks like you shot up 24,000 ppd by going from dual to triple channel - great stuff. :D

What MCH strap are you on?
 
may i be worthy said:
Looks like you shot up 24,000 ppd by going from dual to triple channel - great stuff. :D

What MCH strap are you on?
Click to expand...

1333, 8-8-8-24 1T manually set, everything else(memory wise) left at auto, B2B Cas disabled.

Strangely, disabling NUMA disn't give me any boost with triple channel(it did in dual channel) untill I switched ACPI to v2.0 instead of v3.0, not sure why but... It works now :D
 
Last edited:
@Chilly

6900 0:11:34 156,557
2686 0:11:37 155,554

Are these using the benchmark?
 
may i be worthy said:
@Chilly

6900 0:11:34 156,557
2686 0:11:37 155,554

Are these using the benchmark?
Click to expand...

No, those TPF's taken from WU's I was working on at the time of me tweaking it. I do have results with triple channel with the bench in the chart, but they need to be redone as my testing was flawed(12:07 TPF for 2686 bench). I am guessing my results should be better(closer to my "real" wu performance) once I get around to doing it.
 
Tobit said:
I think it was in the graphics forum or some other forum like it that is constantly filled with too much troll testosterone. :rolleyes:
Click to expand...

That's why I stay in DC and Displays. More "D" less "T"rolls!

EDIT: Added 6900 entry from regular folding and 2684 from the musky Benchmark. Onto 2686!
 
Last edited:
added a plethora of info on a 1090T and Q9450......i wish my 1090T would pick up a 2633 unit here or there, unit is the best one i have for my Q9450 (just about 7.7k ppd on average)
 
W.Feather said:
added a plethora of info on a 1090T and Q9450......i wish my 1090T would pick up a 2633 unit here or there, unit is the best one i have for my Q9450 (just about 7.7k ppd on average)
Click to expand...


1090T cant run the 2633 WU's. they are quad core only.
 
need more phenom II x4 numbers damnit! :D only 2 of us running them, thats sad.
 
sirmonkey1985 said:
need more phenom II x4 numbers damnit! :D only 2 of us running them, thats sad.
Click to expand...

Its probably because they are such a PITA to clock lol. I have oc'ed two AM3 quads and they were so finicky it turned me off from them until at least Bulldozer.

Honestly its likely just the population of folders that are more likely to frequent the forums tend to have hardcore hardware, my quads are pretty pedestrian.
 
R-Type said:
Its probably because they are such a PITA to clock lol. I have oc'ed two AM3 quads and they were so finicky it turned me off from them until at least Bulldozer.

Honestly its likely just the population of folders that are more likely to frequent the forums tend to have hardcore hardware, my quads are pretty pedestrian.
Click to expand...


i find them very easy to overclock. most of the people that have trouple overclocking them are long time intel people since AMD processors dont overclock the same. a lot more things required when overclocking a locked processor and some people have no clue what to do with an unlocked processor.

but then again i guess I'm the exception since I've been overclocking AMD processors since the socket A day's.
 
sirmonkey1985 said:
but then again i guess I'm the exception since I've been overclocking AMD processors since the socket A day's.
Click to expand...
As have I, in fact I used to fold on a 2600m at 2.6ghz back in the day. Your argument is close though.... the Intel chips made me lazy with how easy they are to clock.;)

Meanwhile when the AMD chip passes 12 hours of prime 95 and I set it off to fold and discover it crashed 90% through a WU, I have to resist the urge to upgrade immediately.

I think the x6's are great chips though, and the SMP numbers back that up.
 
or you fail prime 95 after 2 minutes but it does everything else just fine.
why i gave up even using prime95 after my 2500+barton chip.
 
yeah i quit using prime95 too. anytime it passes, F@H fails. so i started using intel burn test which is just a modified GUI for linX stress testing. but so far if i pass that i have absolutely no problems with F@H. plus its a hell of a lot faster. usually just run 10 passes and im good.
 
sirmonkey1985 said:
yeah i quit using prime95 too. anytime it passes, F@H fails. so i started using intel burn test which is just a modified GUI for linX stress testing. but so far if i pass that i have absolutely no problems with F@H. plus its a hell of a lot faster. usually just run 10 passes and im good.
Click to expand...

Good to know, I use IBT as well but never associated it with finding faults quicker. Typically if my Intel chips pass 1-2 hours of prime they're good, it was frustrating having to devote a full day of stress testing to the Athlon anytime I changed a setting.

Today it lost a 6701 unit at around 80% completion because I tightened the ram up to its rated settings and the overclocked northbridge needed more voltage. Now I'm just going to let it be until I upgrade again and it gets bumped out the bottom of my hand-me-down queue.
 
yeah the 6701/2's are picky as shit. but i dont bother with ram timing/speed on standard SMP. makes little to no difference. maybe 1 second in the TPF. where as bigadv it makes a HUGE difference. wouldnt surprise me with the NB voltage. seems they use the absolute minimum stable voltage for the NB stock. wish i could overclock my damn NB though. damn asus motherboard :( im stuck at 1.6Ghz on the NB.
 
sirmonkey1985 said:
yeah the 6701/2's are picky as shit. but i dont bother with ram timing/speed on standard SMP. makes little to no difference. maybe 1 second in the TPF. where as bigadv it makes a HUGE difference. wouldnt surprise me with the NB voltage. seems they use the absolute minimum stable voltage for the NB stock. wish i could overclock my damn NB though. damn asus motherboard :( im stuck at 1.6Ghz on the NB.
Click to expand...

It should oc automatically with your bus clock unless the motherboard is dropping the multiplier on its own. The worst case I have heard is mobos that tie the HT/NB multiplier together.

I have my NB at 2300mhz on stock volts with 9-9-9-34 timings but need to increase the voltage to tighten them up. The Athlon II nb is pretty bad, so I would think even your first spin Phenom 2 should match that.
 
R-Type said:
It should oc automatically with your bus clock unless the motherboard is dropping the multiplier on its own. The worst case I have heard is mobos that tie the HT/NB multiplier together.

I have my NB at 2300mhz on stock volts with 9-9-9-34 timings but need to increase the voltage to tighten them up. The Athlon II nb is pretty bad, so I would think even your first spin Phenom 2 should match that.
Click to expand...


nah its because my board doesnt officially support the phenom II's. has just enough support with these beta bios that it runs and i can overclock it but i have no other controls like NB clock or NB voltage. only things i can change are the cpu voltage, FSB, HTT, cpu multiplier. so im stuck at 1Ghz HTT and 1.6Ghz NB.

the athlon II NB is the same NB in the phenom II 945 C2/940/920. they dont like being overclocked that high and when they do they require a decent amount of voltage. most top out around 2.6Ghz @ 1.35-1.4v on the northbridge.
 
You must log in or register to reply here.
Back
Top