• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

iGPU Folding - Worth It?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Joined
Dec 13, 2005
Location
Right Here Right Now
Hey so my htpc has been sitting idle for a while now. Even when I use it I don't stress it.

I wanted to check if it's worth tossing an F@H client on it. It's got a 5600 in it.

Like a Civic running against Lambos, but if it's worth it I'll toss her in.
 
Not in my experience. At least not on my 7800X3D. Throw a video card in the HTPC and fold on that. Prices on RTX 40 series has been coming down, especially on the used market.
 
I looked at 40xx cards, even 4060s are pushing $250+ on eBay. If there's anything less in the desktop market I'll give it a shot, but this is just about making an idle machine useful.
 
I haven't looked into 4060s but I have found 4070Ti for $500, 4080 super for $750, 4090 for $1100. I'm sure there are deals to be had on 4060s too.
Post magically merged:

A quick marketplace serch brought up 4060Ti in my are from between 250-300. That's asking too. I'm sure you could lowball someone and pick one up for $200.
 
I haven't looked into 4060s but I have found 4070Ti for $500, 4080 super for $750, 4090 for $1100. I'm sure there are deals to be had on 4060s too.
Post magically merged:

A quick marketplace serch brought up 4060Ti in my are from between 250-300. That's asking too. I'm sure you could lowball someone and pick one up for $200.
I've been eyeballing a pulled 4060 that was posted for a c-note. Made an offer for $150.

It's just a basic Dell OEM card. Single fan, PCIe length. Definitely not a gamer card.

If they approve it, yeah it's going to folding.
 
All iGPUs feel like not worth folding and if I'm right it doesn't work on Intel anyway. Compared to discrete GPUs from the last couple of years, they still offer very low performance per W. I don't remember exactly how much it was, but AMD options use 25-50W, while they get 100-200k PPD or something (highly depends on the GPU). In comparison, RTX3060 is around 3M PPD at 150W and RTX4070 is 9M PPD at 150W. Looking at these numbers, RTX3060 feels like a waste of electricity too, but if you have a cheap or free source then why not.
 
Well this would've been the 5600 in my htpc, not the 12700 in my desktop.

But got it going on my 2080 this morning - inefficient as hell going by your numbers, was sitting at an estimated 3.4mill when I left. But I'll figure something out eventually.
 
Well this would've been the 5600 in my htpc, not the 12700 in my desktop.

But got it going on my 2080 this morning - inefficient as hell going by your numbers, was sitting at an estimated 3.4mill when I left. But I'll figure something out eventually.
I get about 3-4m ppd on my 2080's so that's about right.
 
Its worth upgrading to the new stuff. Better PPD per watt. Sell the 2080 for $300 and add another $200. $500 will get you a 4070Ti. They do between 12-17m PPD.


Someone failed the RTX4070 efficiency on that list. I have the FE version and it was 170W / 7-8M PPD at stock settings, and they listed it as 200W / 4.6M PPD. I know the power is listed as TDP, but PPD is still weirdly low. I was running it at 150W / 9M PPD.
RTX4070 Super was 11-17M PPD, depending on WU. I was surprised it kept 15M PPD most of the time when RTX4080 was between 18-24M PPD, but typically closer to 19-20M PPD at 100W more.
 
Lar is an OK list. It gives you a general idea. I always wondered if the sats are messed up from people gaming and turning in a slow WU.

My 4080 Super is dong 19-24m on average at 300w, 4070 super 220w, 4070TI 270w
 
Lar is an OK list. It gives you a general idea. I always wondered if the sats are messed up from people gaming and turning in a slow WU.

My 4080 Super is dong 19-24m on average at 300w, 4070 super 220w, 4070TI 270w

For RTX4070 Super and RTX4080, it's about correct looking at my results, but for RTX4070 it's far from reality as after some overclocking I was reaching 100% more. On RTX3060 I also had almost 1M more at nothing but auto/stock settings and it's quite a big difference if they say it is supposed to have 2.6M and I had ~3.5M PPD. Maybe some people run additional stuff in the background or keep the web browser open, I don't know.
 
For RTX4070 Super and RTX4080, it's about correct looking at my results, but for RTX4070 it's far from reality as after some overclocking I was reaching 100% more. On RTX3060 I also had almost 1M more at nothing but auto/stock settings and it's quite a big difference if they say it is supposed to have 2.6M and I had ~3.5M PPD. Maybe some people run additional stuff in the background or keep the web browser open, I don't know.
Yeah, Lar is very hit or miss. There are some cards that are pretty accurate, but I've found that most are quite a bit off. It also doesn't take into account platform differences which can be quite huge with the newer cards. We've seen up to 4ish million ppd per card differences between machines running the same cards. PCIE bandwidth and CPU power (for checkpoints) are big, ram speeds to a lesser extent. Multi-card boxes do quite a bit better in Linux than Windows. Like @KeeperOfTheButch said, lars is good for a general idea, but small things like reboots in the middle of a WU, pausing, gaming with f@h running, etc can skew the results significantly. Heck, I had one machine taking 30sec to a minute to upload the WU after it was finished (one time it took 5min) and that was enough to tank ppd by like a .5-1m points.
 
Back