Page 3 of 5
Re: GTX 1660Ti Compute Performance
Posted: Wed Mar 06, 2019 3:31 pm
by jjbduke2004
1st update!
Card type: GIGABYTE GeForce GTX 1660 Ti WINDFORCE OC 6G
Card speed: 1845 MHz (I have not increased beyond the mfgr's overclock)
It looks like it's capable of around 700k PPD. However, I was watching quite a few youtube streams in the evening (8 railroad web cams tiled on a 2560x1440 display) so that consumes video card resources. I believe the HTML5 player makes use of the GPU processing power. That dropped me down to the 600k range. I'm at work and it's still running (without youtube going) so we'll see what happens.
If you want to follow my stats, my Folding name is the same as my forum name.
Re: GTX 1660Ti Compute Performance
Posted: Wed Mar 06, 2019 4:04 pm
by gordonbb
jjbduke2004 wrote:1st update!
Card type: GIGABYTE GeForce GTX 1660 Ti WINDFORCE OC 6G
Card speed: 1845 MHz (I have not increased beyond the mfgr's overclock)
It looks like it's capable of around 700k PPD. However, I was watching quite a few youtube streams in the evening (8 railroad web cams tiled on a 2560x1440 display) so that consumes video card resources. I believe the HTML5 player makes use of the GPU processing power. That dropped me down to the 600k range. I'm at work and it's still running (without youtube going) so we'll see what happens.
If you want to follow my stats, my Folding name is the same as my forum name.
Thanks,
Can you let us know what OS you are running as well please?
Re: GTX 1660Ti Compute Performance
Posted: Wed Mar 06, 2019 4:26 pm
by jjbduke2004
gordonbb wrote:
Can you let us know what OS you are running as well please?
Windows 10 64-bit
Edit: I'll also add that my CPU is an i5-3570K and I'm not running any of the CPU cores.
Re: GTX 1660Ti Compute Performance
Posted: Wed Mar 06, 2019 5:42 pm
by bruce
jjbduke2004 wrote: I'm not running any of the CPU cores.
The general consensus is that concurrent folding with CPU cores adds a relatively few PPD but it's best to leave some free CPU resources so that the GPU can use whatever it wants whenever it wants to.
Re: GTX 1660Ti Compute Performance
Posted: Wed Mar 06, 2019 9:17 pm
by toTOW
MeeLee wrote:toTOW wrote:None of the new TU116 variants were supported by FAH. I just added them.
How do you add them, without the card? Or did you get yourself one?
I don't need the cards to find the PCI IDs ...
Re: GTX 1660Ti Compute Performance
Posted: Fri Mar 08, 2019 5:44 am
by MeeLee
jjbduke2004 wrote:1st update!
Card type: GIGABYTE GeForce GTX 1660 Ti WINDFORCE OC 6G
Card speed: 1845 MHz (I have not increased beyond the mfgr's overclock)
It looks like it's capable of around 700k PPD. However, I was watching quite a few youtube streams in the evening (8 railroad web cams tiled on a 2560x1440 display) so that consumes video card resources. I believe the HTML5 player makes use of the GPU processing power. That dropped me down to the 600k range. I'm at work and it's still running (without youtube going) so we'll see what happens.
If you want to follow my stats, my Folding name is the same as my forum name.
700k sounds about right.
It'll be more when you have no screensaver, and let the PC run it's course.
My intitial estimate would have been 750k PPD.
Re: GTX 1660Ti Compute Performance
Posted: Fri Mar 08, 2019 4:52 pm
by jjbduke2004
Update 2:
The PPD estimator in Advanced Controls generally hovers around 630k PPD. Obviously, different project's work units produce different results. For the two full calendar days that I've been folding, I haven't broken 600k (528k and 594k). In actuality, a large 108,800 unit was submitted like 2 minutes before midnight (CT) last night, but did not make the midnight stats run.
Looking at rolling 24 hour windows (and going by the current stats run as far back as 9pm March 6):
24 hours ending 9pm March 7: 763339 points
24 hours ending 6pm March 7: 653303 points
To add, my screen saver is "blank screen."
I've definitely noticed an increase in TPF when actively browsing the web. I'll give it a few more days of close-to-dedicated folding, before I resume normal PC activities (watching youtube, playing Wolfenstein II, etc.). I'll continue with FAH in the background a little bit longer.
Re: GTX 1660Ti Compute Performance
Posted: Sat Mar 09, 2019 5:46 am
by Theodore
Using your PC while folding does affect PPDs in a negative way.
This is why most people post their PPD ratings, when the system is idle; it's their 'highest' PPD they post; not averages from power outages or system maintenance, or while using the PC.
Re: GTX 1660Ti Compute Performance
Posted: Sat Mar 09, 2019 8:00 am
by Nathan_P
From jjbduke2004's numbers it looks to be the same as a 1070, not bad at all.
Re: GTX 1660Ti Compute Performance
Posted: Sat Mar 09, 2019 7:47 pm
by bruce
Nathan_P wrote:...it looks to be the same as a 1070, not bad at all.
The 1070 produces 5783 (6463) single precision GFLOPS. The 1660Ti produces 4608 (5437) SP GFLOPS. The 1660Ti has a significantly faster memory subsystem.
Re: GTX 1660Ti Compute Performance
Posted: Sat Mar 09, 2019 11:20 pm
by toTOW
In the GPGPU world, not all FLOPs are equal ...
Re: GTX 1660Ti Compute Performance
Posted: Sun Mar 10, 2019 1:23 am
by bruce
Explain ... with examples.
Re: GTX 1660Ti Compute Performance
Posted: Sun Mar 10, 2019 12:24 pm
by MeeLee
There are rumors about Nvidia releasing a cut-down version of the 1660 ti; the 1660, and possibly even a 1650; all cut down on cuda cores, shaders, etc...
If the 1660 ti only gets 600k PPD, it won't be very interesting compared to the larger RTX 2060 model.
Re: GTX 1660Ti Compute Performance
Posted: Mon Mar 11, 2019 2:36 pm
by jjbduke2004
Update 3:
3/8 closed out with 689,896 points.
The last bunch of work units I've been getting are not as friendly so my PPD is below 500k.
So the card is capable of 700k+ peak but YMMV. I hope this helps anyone considering a purchase.
I will be returning to normal PC operations, but probably fold a little here and there before the summer. If you have specific questions or want more details from the logs regarding PPD for differing project types, PM me.
Re: GTX 1660Ti Compute Performance
Posted: Tue Mar 12, 2019 3:26 pm
by jjbduke2004
Update 4: I think something happened midway through that caused the card to stop "Boosting" and run at stock 1500 MHz despite wiggle room in heat dissipation and power consumption. I upgraded drivers as a new release just came out (419.17 -> 419.35) and the reload seemed to help activate boost (either that or I mucked with the right things in Aorus Engine). Anyway the card is now running between 1905 and 1920 MHz (though I haven't raised the boost speed above 1845, it seems to do it's own thing, which I've learned is "normal" for NVIDIA Boost 3.0/4.0).
I do notice that the graphics core is taking up about 30% CPU which I imagine means that one of the four processor cores is fully occupied with another running a second, less demanding thread. Video bus usage (PCIe 3.0) was around 43% so I wonder if my processor is the bottleneck in feeding the data since the past few work units only occupied 87-93% of the video card.