Some questions about the Geforce GT640

Message boards : Number crunching : Some questions about the Geforce GT640
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · Next

AuthorMessage
Profile ivan
Volunteer tester
Avatar

Send message
Joined: 5 Mar 01
Posts: 783
Credit: 348,560,338
RAC: 223
United Kingdom
Message 1464087 - Posted: 13 Jan 2014, 22:35:12 UTC - in response to Message 1463943.  

I'm thinking of getting one of these:
GT640DDR5 95°C MaxTemp 48W Max Graphics Card Power 384CUDA cores

Specifically, a particular Gigabyte model that now seems to be out of stock at eBuyer. Hmm, on pre-order at Scan. Reason is, I have a spare PCIE-16 slot on my new motherboard but it's only one slot wide, so any double-width board won't fit. I'm just a little bit doubtful as it's very close to the bottom of the PC case and I wonder if it will get enough airflow through the fan. Still, it's only 49 W so it shouldn't need a great amount of cooling.
ID: 1464087 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13727
Credit: 208,696,464
RAC: 304
Australia
Message 1464092 - Posted: 13 Jan 2014, 22:44:30 UTC - in response to Message 1464087.  
Last modified: 13 Jan 2014, 22:46:41 UTC

Reason is, I have a spare PCIE-16 slot on my new motherboard but it's only one slot wide, so any double-width board won't fit.

Double check that it is one slot width wide- the standard model is double slot width, but many of the manufacturers do their own heatsink & fans on the models they release.

EDIT- checked your link and it is 1 slot width wide. Given the low power of the card I doubt you'd have any cooling issues, especially with ambient temperatures well below 25°c most of the time.
Grant
Darwin NT
ID: 1464092 · Report as offensive
Profile ivan
Volunteer tester
Avatar

Send message
Joined: 5 Mar 01
Posts: 783
Credit: 348,560,338
RAC: 223
United Kingdom
Message 1464096 - Posted: 13 Jan 2014, 23:12:24 UTC - in response to Message 1464092.  

Reason is, I have a spare PCIE-16 slot on my new motherboard but it's only one slot wide, so any double-width board won't fit.

Double check that it is one slot width wide- the standard model is double slot width, but many of the manufacturers do their own heatsink & fans on the models they release.

EDIT- checked your link and it is 1 slot width wide. Given the low power of the card I doubt you'd have any cooling issues, especially with ambient temperatures well below 25°c most of the time.

Yeah, it's not Darwin, but it can get up to 25 C or more in summer here. It'd be nice to push the home PC along a bit in RAC -- the DDR5 GT 640 has a surprising FLOPS rating for the price and power. I'm tempted to swap one into my work PC too, as it has a proprietary [(hp)] power supply which doesn't have GPU power connectors (and physically can't be replaced by a standard PS).
ID: 1464096 · Report as offensive
Profile Jeff Buck Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester

Send message
Joined: 11 Feb 00
Posts: 1441
Credit: 148,764,870
RAC: 0
United States
Message 1464097 - Posted: 13 Jan 2014, 23:20:06 UTC - in response to Message 1464087.  

it's only one slot wide, so any double-width board won't fit.

Even though it's technically only one slot wide, you want to make sure that there's nothing plugged into those sockets at the edge of the board that might prevent the card from seating all the way. I had that problem when I first tried to put a GT640 into my IBM ThinkCentre. The way the motherboard is configured, the socket for the motherboard's main power plug is right next to the PCIe slot and the boxy housing for the GT640's cooling fan was trying to occupy a tiny bit of the same space that the main power plug was already using. I ended up having to use a riser cable to lift the GT640 off the board. The upside of that experience, though, was that I learned I could use riser cables to make use of several otherwise inaccessible PCIe slots on two of my other machines.
ID: 1464097 · Report as offensive
Profile ivan
Volunteer tester
Avatar

Send message
Joined: 5 Mar 01
Posts: 783
Credit: 348,560,338
RAC: 223
United Kingdom
Message 1464120 - Posted: 14 Jan 2014, 0:57:47 UTC - in response to Message 1464097.  

it's only one slot wide, so any double-width board won't fit.

Even though it's technically only one slot wide, you want to make sure that there's nothing plugged into those sockets at the edge of the board that might prevent the card from seating all the way.

Good point. There's a front-panel audio and front-panel USB headers that might interfere. OTOH, I've plenty of other machines I can swap it to if it proves problematic.
ID: 1464120 · Report as offensive
_
Avatar

Send message
Joined: 15 Nov 12
Posts: 299
Credit: 9,037,618
RAC: 0
United States
Message 1464148 - Posted: 14 Jan 2014, 4:38:34 UTC

So I bit the bullet and ordered two of these guys: ASUS 2GB GTX 650ti

I know that some have mentioned the 2GB is unnecessary, but it wasn't that much more expensive and if I ever need a card for my main (non-seti) rig, I'll have it.

While I am waiting, I have been thinking of what I can do in the mean time before the cards arrive. These are some preliminary thoughts:

1)Of course let my current work units finish.

2)It seems like it would be a good time to upgrade BOINC. So I will do that.

3)Should I uninstall any drivers for my old cards? The old cards are NVIDIA GeForce 8800 GTS 512 (512MB) -and- AMD AMD Radeon HD 7700 series (Cape Verde) (1024MB).

4)Once the cards come, I'll rerun the lunatics installer and likely choose cuda50. (Thanks to Phil and his thread Located Here

5)I will attempt to run 2 WU's per card. MB at first. If I get cocky I'll try to figure out simultaneous AP.

6)Change the GPU priority to above normal in mbcuda.cfg ?? We will see about this one.

Hopefully this is a good list to reference once my cards actually come.

Thanks again everyone for guiding me towards a great purchase!
ID: 1464148 · Report as offensive
Philhnnss
Volunteer tester

Send message
Joined: 22 Feb 08
Posts: 63
Credit: 30,694,327
RAC: 162
United States
Message 1464190 - Posted: 14 Jan 2014, 6:22:00 UTC - in response to Message 1464010.  
Last modified: 14 Jan 2014, 6:25:19 UTC

Yes, according to my statis page Bonic has ran 94.1905 % of the time the
computer is running, you got me currious? So I would say that's pretty
close to 24/7, LOL!!!

I have not changed the GPU setting to above normal on that computer yet. I
really want to see what my RAC will level off to first. From the rate it is
still climbing I am guessing about two more weeks. It just surprised the heck
out of me what a big differance the changes I did make made. I guess I want
to be surprised again!!

I ran a pair of 250's for a couple of years before I got the two 450's.
You'll be amazed at how much cooler your computer will run. I know I
had to under clock the 250's just to manage the heat. In any case I hope
your as happy with the two 650 TI's as I am after I followed the great advice
from everybody!!!
ID: 1464190 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22184
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1464195 - Posted: 14 Jan 2014, 6:44:55 UTC

Given the big step in GPU family (for the Nvidia card) I would remove the old drivers just to make sure you get a clean run at things. If possible do this with no GPUs installed, then install fresh from Nvidia- avoid the MS offering like the plague!!!!
Start with one WU per GPU for a couple of days to make sure you haven't got a problem card and keep a very close eye on the results for the first hour or so.
Don't use SLI as SETI doesn't use it.
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1464195 · Report as offensive
_
Avatar

Send message
Joined: 15 Nov 12
Posts: 299
Credit: 9,037,618
RAC: 0
United States
Message 1464587 - Posted: 15 Jan 2014, 13:47:25 UTC - in response to Message 1464190.  

Yes, according to my statis page Bonic has ran 94.1905 % of the time the
computer is running, you got me currious? So I would say that's pretty
close to 24/7, LOL!!!

I have not changed the GPU setting to above normal on that computer yet. I
really want to see what my RAC will level off to first. From the rate it is
still climbing I am guessing about two more weeks. It just surprised the heck
out of me what a big differance the changes I did make made. I guess I want
to be surprised again!!

I ran a pair of 250's for a couple of years before I got the two 450's.
You'll be amazed at how much cooler your computer will run. I know I
had to under clock the 250's just to manage the heat. In any case I hope
your as happy with the two 650 TI's as I am after I followed the great advice
from everybody!!!



Thank you very much Phil for your input on this topic. You are exactly in the situation I will soon be in, so it is really great to hear someone speak from experience on these cards. Thanks a lot for that.

And thank you too Rob, I will try to uninstall whatever drivers I can find. I'm sure the cards will come with their own drivers, I wonder if they will be sufficient or if I will have to venture to the NVIDIA site for the latest.

The cards should be coming today!
ID: 1464587 · Report as offensive
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 9786
Credit: 572,710,851
RAC: 3,799
Panama
Message 1464590 - Posted: 15 Jan 2014, 13:53:30 UTC - in response to Message 1464587.  
Last modified: 15 Jan 2014, 13:57:07 UTC

I wonder if they will be sufficient or if I will have to venture to the NVIDIA site for the latest.

The driver that comes with the board normaly is old. Go to the NVidia site, DL the latest WHQL driver for your GPU. And when you do the instalation ALLWAYS REMEMBER, install from CLEAN INSTALATION and never allow Windows Update update your video driver. This will avoid you a lot of troubles and headaches and a lot of hours of happy crunching.

Another Tip: Keep an eye on your GPU temperature, use Precision-x, Afterburner or any similar program who allow you to increase the fan speed and mantain your GPU into a safe temperature range (varies from one GPU model to the other).
ID: 1464590 · Report as offensive
_
Avatar

Send message
Joined: 15 Nov 12
Posts: 299
Credit: 9,037,618
RAC: 0
United States
Message 1464595 - Posted: 15 Jan 2014, 14:12:58 UTC - in response to Message 1464590.  
Last modified: 15 Jan 2014, 14:14:25 UTC

I wonder if they will be sufficient or if I will have to venture to the NVIDIA site for the latest.

The driver that comes with the board normaly is old. Go to the NVidia site, DL the latest WHQL driver for your GPU. And when you do the instalation ALLWAYS REMEMBER, install from CLEAN INSTALATION and never allow Windows Update update your video driver. This will avoid you a lot of troubles and headaches and a lot of hours of happy crunching.

Another Tip: Keep an eye on your GPU temperature, use Precision-x, Afterburner or any similar program who allow you to increase the fan speed and mantain your GPU into a safe temperature range (varies from one GPU model to the other).


Thanks for the advice, Juan. By clean installation, do you mean to find the drivers myself and manually run their installation files? Instead of the windows "searching for drivers..." wizard that you can sometimes run into?

As far as the GPU temperature, I think I will start off slow with 1 WU per card and see what happens. Rob suggested that I do this to make sure the results are at least valid from this. After that, I will monitor the temperature when I start to try 2 WU at a time. Leo stated that the max temp for the cards I've purchased it 105°C. He says he even can run 3 or 4 WUs on his cards. I guess I will see if that is something I should do when I monitor the GPU usage.

I suppose it is a fine line to walk? On one hand, you are aiming for your GPU to be doing 95% of its capacity. At the same time, you don't want it to overheat. One a day where I can really take some time to monitor the temperatures, I will do just that. I have never seen a program that will allow you to speed up your GPU fan, so that is a great idea to look into.
ID: 1464595 · Report as offensive
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 9786
Credit: 572,710,851
RAC: 3,799
Panama
Message 1464599 - Posted: 15 Jan 2014, 14:32:04 UTC - in response to Message 1464595.  
Last modified: 15 Jan 2014, 14:44:37 UTC

Thanks for the advice, Juan. By clean installation, do you mean to find the drivers myself and manually run their installation files? Instead of the windows "searching for drivers..." wizard that you can sometimes run into?

No, what i say is not allow the Windows Update Program install/update your video driver, not the windows basic boot system (not sure what is the real name). At the boot the windows search for the basic driver when you install any new device (any device GPU, HD, SDD, etc), leave it install the basic device driver, or your GPU will not be recognized as a new device for the windows itself. After that DL the driver from Nvidia, and run the Nvidia setup program (just open the DL file). It will prompt you to choose several options to install, choose your destination of the instalation files and another windows will open with more options that allow you to choose what software you want to install, at the end of the list, on the left corner you will see: clean instalation, besure to click there and the program will do the rest. Do the same each time you need to upgrade your driver, that will guaranted you your instalation program will clean first whatever the old program leaves and allow a compleate new instalation - free of bug (or at least is what we expect form a new driver - not allways true of course)

As far as the GPU temperature, I think I will start off slow with 1 WU per card and see what happens. Rob suggested that I do this to make sure the results are at least valid from this. After that, I will monitor the temperature when I start to try 2 WU at a time. Leo stated that the max temp for the cards I've purchased it 105°C. He says he even can run 3 or 4 WUs on his cards. I guess I will see if that is something I should do when I monitor the GPU usage.

I suppose it is a fine line to walk? On one hand, you are aiming for your GPU to be doing 95% of its capacity. At the same time, you don't want it to overheat. One a day where I can really take some time to monitor the temperatures, I will do just that. I have never seen a program that will allow you to speed up your GPU fan, so that is a great idea to look into.

Rob is right and this is the right path to follow, my sugestion is to use a fan controler program even if you run only 1 WU at a time, i run 2 on most of my GPU´s (only run 3 on the 780´s) and belive it´s the optimal setting for your too. Remember even if your GPu could run more WU at a time there is an optimal setting, in my case 2 is better than one, 3 give an aditional gain on the 780 and 4 is worst, but each system is unike so the roule YMMV is in place. I don´t have the same GPU you post, i realy belive 2 is the best setting for it but only you could be sure about that.

High temperatures is allways the enemy of the GPU´s. The stock setting is target for games and not for crunching, so if you want to do some crunching, increase the fan speed is a good practice. I use EVGA Precision-X to do that, but there are normaly a similar program avaiable form your GPU Site, if no try the EVGA program, it´s free, easy to use and works perfect on allmost any Nvidia GPU. Try to mantain the temp lowers as possible, that will allow you GPu runs faster and extend it´s life time. And not forget, often clean the fan and air paths from dust, it´s incredible the quantity of dust acumulated in few months of work.

Happy crunching
ID: 1464599 · Report as offensive
_
Avatar

Send message
Joined: 15 Nov 12
Posts: 299
Credit: 9,037,618
RAC: 0
United States
Message 1464608 - Posted: 15 Jan 2014, 15:12:50 UTC - in response to Message 1464599.  

This is great advice all around Juan. Thank you very much. You were also giving me great advice when I initially set up some GPU's a few months ago. Thanks again for giving great information this time, too!
ID: 1464608 · Report as offensive
_
Avatar

Send message
Joined: 15 Nov 12
Posts: 299
Credit: 9,037,618
RAC: 0
United States
Message 1464616 - Posted: 15 Jan 2014, 15:42:15 UTC

A random question for anyone who still may be peeking at this thread:

If I have some CPU tasks still waiting to finish up, but all of my GPU tasks are finished, would it be safe to reinstall BOINC and replace the GPU cards?

I am just not completely clear whether the CPU tasks would still be there after a BOINC version upgrade.
ID: 1464616 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14650
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1464619 - Posted: 15 Jan 2014, 15:47:32 UTC - in response to Message 1464616.  

A random question for anyone who still may be peeking at this thread:

If I have some CPU tasks still waiting to finish up, but all of my GPU tasks are finished, would it be safe to reinstall BOINC and replace the GPU cards?

I am just not completely clear whether the CPU tasks would still be there after a BOINC version upgrade.

It should be perfectly safe. It's possible for things to go wrong, as with everything in life, and it's possible to screw things up yourself (but you have to work quite hard to achieve that). I upgrade several machines to pretty much every version of BOINC that's put up for testing, and I've given up taking any precautions - they just work.
ID: 1464619 · Report as offensive
_
Avatar

Send message
Joined: 15 Nov 12
Posts: 299
Credit: 9,037,618
RAC: 0
United States
Message 1464621 - Posted: 15 Jan 2014, 15:49:14 UTC - in response to Message 1464619.  

Great! This means I won't have to wait another day or two to install the new cards!

Thanks Richard!
ID: 1464621 · Report as offensive
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 9786
Credit: 572,710,851
RAC: 3,799
Panama
Message 1464631 - Posted: 15 Jan 2014, 16:30:27 UTC

Since your hosts are hidden we can´t see what Boinc you actualy ussing. You must see if this applies to you if you want to upgrade the Boinc itself:

From http://boinc.berkeley.edu/wiki/Release_Notes

Known issues with BOINC 7.2.26 and Seti when updating from BOINC 7.0.28 or before and getting 197 (0xc5) EXIT_TIME_LIMIT_EXCEEDED errors on anonymous platform GPU: When you've updated to BOINC 7.2.26, coming from BOINC 7.0.28 or before, AND when crunching work on an anonymous platform GPU application (such as the ones from Lunatics), you have a good chance that all your work errs with a 197 (0xc5) EXIT_TIME_LIMIT_EXCEEDED error.
This is because of an addition to the code in BOINC 7.0.33, where when you run the anonymous platform and you have not supplied a flops value in your app_info.xml file, that the flops value used for GPU tasks is increased by a factor of 10.
It will only affect old work still in cache from before the upgrade. Any newer work your BOINC downloads, will not be affected.
Two ways to work around this:
1. Before you upgrade, set No New Tasks and run your cache empty. Report all. Then upgrade to the new version and set Allow New Tasks.
2. Done the upgrade with a full cache? Reset the project. This will throw away all present work in cache, but since the project has resend lost work available, your tasks will be resent to you and seen as new work.

ID: 1464631 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14650
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1464637 - Posted: 15 Jan 2014, 16:42:27 UTC - in response to Message 1464631.  

Since your hosts are hidden we can´t see what Boinc you actualy ussing. You must see if this applies to you if you want to upgrade the Boinc itself:

From http://boinc.berkeley.edu/wiki/Release_Notes

Known issues with BOINC 7.2.26 and Seti when updating from BOINC 7.0.28 or before and getting 197 (0xc5) EXIT_TIME_LIMIT_EXCEEDED errors on anonymous platform GPU: When you've updated to BOINC 7.2.26, coming from BOINC 7.0.28 or before, AND when crunching work on an anonymous platform GPU application (such as the ones from Lunatics), you have a good chance that all your work errs with a 197 (0xc5) EXIT_TIME_LIMIT_EXCEEDED error.
This is because of an addition to the code in BOINC 7.0.33, where when you run the anonymous platform and you have not supplied a flops value in your app_info.xml file, that the flops value used for GPU tasks is increased by a factor of 10.
It will only affect old work still in cache from before the upgrade. Any newer work your BOINC downloads, will not be affected.
Two ways to work around this:
1. Before you upgrade, set No New Tasks and run your cache empty. Report all. Then upgrade to the new version and set Allow New Tasks.
2. Done the upgrade with a full cache? Reset the project. This will throw away all present work in cache, but since the project has resend lost work available, your tasks will be resent to you and seen as new work.

That's a wise warning in general (which is why we also have it in a sticky thread here). But it only applies to GPU tasks, and mherr has told us he's finished all the GPU tasks already: he should be good to go.
ID: 1464637 · Report as offensive
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 9786
Credit: 572,710,851
RAC: 3,799
Panama
Message 1464663 - Posted: 15 Jan 2014, 17:45:58 UTC - in response to Message 1464637.  
Last modified: 15 Jan 2014, 17:47:19 UTC

That's a wise warning in general (which is why we also have it in a sticky thread here). But it only applies to GPU tasks, and mherr has told us he's finished all the GPU tasks already: he should be good to go.

You are right as allways Richard. That´s why i said: You must see if this applies to you since his host are hidden we can´t be sure all GPU task are all ready realy done in all his hosts (if he have more than one). :)

BTW: Someone who has admin rights could start a fix thread asking those who ask for help first unhidde their hosts so we don´t waste time to point something that is all ready done or maybe point in a wrong direction.
ID: 1464663 · Report as offensive
Philhnnss
Volunteer tester

Send message
Joined: 22 Feb 08
Posts: 63
Credit: 30,694,327
RAC: 162
United States
Message 1464675 - Posted: 15 Jan 2014, 17:57:39 UTC - in response to Message 1464637.  
Last modified: 15 Jan 2014, 18:05:46 UTC

Those ASUS cards will come with a program called GPU Tweak, at least mine did
and they are the same cards. well with the exception of mine just has the 1
GIG of memory. It's not the greatest program but it will work. Since I am
still running Vista I am used to the, not the greatest but it will work.
Using that program, I just set my fans to run at 75% all the time and forgot
about them. Being dual fan cards they make less noise than single fan cards.
And above 75 % I could hear them over the computer with the two 450's. They
normally run around 50 C to 52 C depending on the room tempature. With the
bottem card allways around 1 degree C higher. But I also use those Coolmaster
High AirFlow full size cases. So you may have to adjust according to your
situation.

If like me you can not run CUDA 5.5 here is the link to CUDA 5.035 from
NVIDIA's archive site,

https://developer.nvidia.com/cuda-toolkit-50-archive

it will come with a driver. Won't be the newest one out, so you will want to
run the video update and install the latest driver after you've installed CUDA.


Thank everybody that helped me in that thread. I just followed what was
recomended with FANTASTIC results!!!!!
ID: 1464675 · Report as offensive
Previous · 1 · 2 · 3 · Next

Message boards : Number crunching : Some questions about the Geforce GT640


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.