• 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.

GPU + SMP on Dual Core?

phorensic

n00b
Joined
Jun 18, 2003
Messages
61
I'm trying to max my PPD, so here is a question I have been wondering for a while, but too lazy to experiment.

Windows XP x64, 9600GT, Core2Duo @ 3.5GHz. Running GPU client and a single console client (CPU). The GPU client pulls 100% of core 2, so I run a non-smp console client to fill up the other core. I understand that under Vista the GPU client will not pull 100% of a core, so you can run the SMP client to take up the slack. Well, since this is XP-x64, what would happen if you ran the SMP client on both cores plus the GPU client? How would you want to configure it and would the PPD increase?

 
You should get a ppd decrease for you GPU and a some points for your SMP. I don't know how much for sure, but it is worth a try. For a comparable system. I have an E6700@3.2 and an 8800GT. The 8800GT is getting about 5400 ppd and the SMP is getting about ~1500 ppd, but this is on 64 bit Vista.
 
Does anyone have a technical reason why the GPU client takes 100% of a core on XP but significantly less on Vista? Makes me think that it's saturating that one core with "fake" usage due to a quirk in the programming. I mean, the GPU's are putting out similar PPD on either Vista or XP, but XP saturates a whole CPU core to do it. Seems like you could suck up some of that "fake" usage with the SMP client.
 
Does anyone have a technical reason why the GPU client takes 100% of a core on XP but significantly less on Vista? Makes me think that it's saturating that one core with "fake" usage due to a quirk in the programming. I mean, the GPU's are putting out similar PPD on either Vista or XP, but XP saturates a whole CPU core to do it. Seems like you could suck up some of that "fake" usage with the SMP client.

You can run a SMP+GPU without too much problems, it's just some difference about cpu load calculations as you assumed.

 
I have two systems both amd phenom quad cores (vista 32). The first system is running a pair of HD4870 and the cpu load is 100%. The second system is running four HD4850s and the cpu load is 100%. As i understand the ati client still uses a lot of cpu, correct? When i was setting up the first system 2 weeks ago i tried loading the SMP client but after 20 mins, it still showed working but not even 1% was done.

Am i correct in assuming that GPU and SMP should only be ran in an intel/nvidia setup? or is it ok if i load the SMP client on my system? If the SMP client takes more cpu and slows down the cards would the point offset still be higher, or would i lose ppd ?

 
I've been playing around with the GPU clients. I have one running on an ATI 3850 and the other running on an 8600GT. I have noticed that the ATI client saturates a full core, while the NVIDIA client is barely a blip.
 
I went from 4000ppd+ on the GPU and 200ppd+ on the unused CPU core to... 1600-1700ppd on the SMP and GPU clients. What's weird is that the points are almost exactly the same for the SMP client vs the GPU client. Overall, I dropped from around 4200ppd to around 3200ppd...no go. :(

Oh, I also upped the priority on the GPU client. And looking at task manager shows the GPU core pulling 2-5% CPU instead of 50% (100% of a core).

My conclusion? Buy a quad core, lol.
 
It seems I jumped the gun a bit on performance numbers. After a few more days of folding I am measuring about 2800-3000ppd on the GPU and around 1600ppd on the SMP client. This increase in performance is directly related to the fact that now, all of a sudden, the GPU client is stealing more of a cpu core, around 10-20% usage in task manager. So yeah, at least under WinXP, the GPU client is CPU bottlenecked. And somehow, I'm showing a net gain in PPD now....hmmm, interesante.
 
Did your GPU client upload the new v1.15 core last night ??
If it did then the new speed increase could be related to that.
Check your log file for when the protien start, it will show you what version core you are useing.

Luck .......... :D
 
I manually updated to 1.15 core a few days before beginning this experiment, so that rules out that boost.
 
Back
Top