Do it! I'm kinda glad I took the plunge myself. I game at 4K, and the 2070 Super was okay if you turn down the IQ settings. With the 2080 Ti, this thing can run a solid 60 FPS between high and ultra IQ. I'm actually surprised at how well my 2600K has held up, but then again, at 4K, the GPU is the bottleneck the vast majority of the time.
On another note, my 2080 Ti downclocked itself and was running at 1350 MHz a good part of today, so will take a hit on PPD. Had to reboot the computer to resolve the low clock.
Ha! A 2600k and a 2080Ti! Magnificent!
I need to update my sig, but I've moved on from my hexacore Sandybridge... to a hexacore Ryzen 3. I really have only played Borderlands 3 on it (we don't need to say how many hours), but wow is it so much smoother and there lack of FPS dips and stutters is amazing, not to mention the load times are cut hard between the new hardware and the NVME drive. The 5700 xt didn't change, but my perception of its 2k performance certainly did.
Personally, I think we are close enough to the 3000 series to wait until fall and get one of those. I’ll be upgrading CPUs and playing with some cheaper hardware in the mean time.
Have you ever looked into whether you could harvest the heat to, say partially heat your hot water? That would also reduce the heat load in your house, maybe something as relatively 'simple' as passing the hot air through a radiator (I'm thinking car style here, not house) in the water supply to the hot water tank/boiler etc. Quite how you would duct all, or even a large portion of that hot air into one place is another matter though! Unless the waste heat is already ducted??I own 2 houses outright, no mortgage, one is a rental.
Home solar cost 24k for one house. And wouldn't cover more than $100 a month of my electric. CPUs are now taking some heat, the EPYC boxes do get a little warm (64 cores/128 threads) But a 1070TI produces as much heat as those.
Now if I had an acre or two for a solar farm, that's a different story.
A certain drop of daily production for both teams, but also for other teams as far as I have seen. I am not sure if this is due to work supply or just lagging stats exports again.
I'm guessing work supply. Mine were running solid for a while and then late yesterday/early today I have had to pause and start because they were up to waiting 30 minutes to retry for work. Both are client type advanced as well.
3 of my 9 GPUs are out of service now: The triple-GPU computer shut itself off. On power-on, the BIOS complains about CPU overtemperature and lets me enter the BIOS setup (where it shows 89 °C CPU temperature), but doesn't let me boot into the OS. So I can't at least clear its current F@h and Rosetta work caches. Except perhaps if I plug the disk into another computer.
The pump is working, the water is cool, so I suspect the CPU waterblock is blocked with old dirt in the loop. I will have to take it apart and open the CPU and GPU waterblocks to check.
Couldn't have this happened before the weekend rather than on Sunday evening?
Today being a holiday, I disassembled the loop and opened each of the four waterblocks. The CPU block, which has narrower fins than the GPU blocks, was completely blocked by gunk, as I expected. The first and second GPU block were almost clean. The third GPU block had a lot of gunk in it. That blob of gunk almost jumped at me the moment I opened the waterblock. =:-OTo add to this, I have got four watercooled PCs, all left without maintenance since winter 2017/2018 except for blowing dust out now and then. The main difference of the 3-GPU computer vs. the other ones is that it contains components of my first watercooled PC from 2016. It had a modular all-in-one cooling, which had sealant on the fittings put on by the manufacturer. The sealant had already worked itself into the loop some time in 2017. My suspicion is that there are still remains of this somewhere in this loop. What I am trying to say is that watercooling can work reliably and even maintenance-free until the computing components become obsolete, but I took a risk with re-using components from an earlier failed loop.
No work here. No Work No Production.Day 23: Friday, April 10
14:00 UTC stats
(10 AM EDT, 7 AM PDT)
links: official race stats, TeAm AnandTech at EOC, Tom's Hardware at EOC
Daily production
View attachment 19426
Our advantage
View attachment 19427
A certain drop of daily production for both teams, but also for other teams as far as I have seen. I am not sure if this is due to work supply or just lagging stats exports again.
No work here. No Work No Production.
In case that you haven't already seen it somewhere buried in this thread: If the client doesn't get work at a request, it retries after a period which gets longer after each unsuccessful request. Meaning, the number of requests in a given time frame gets fewer and fewer. In the advanced control (FahControl), the retry period can be seen when you click on the work unit which is in "downloading" state, on the right as "Next Attempt".No work here. No Work No Production.
This is the best resource I know of:What is the radeon vs. nvidia performance situation ?
I can't seem to find a good benchmark chart comparing all the current cards on the market.
I'm guessing because they are partnered with nvidia, their software is mostly optimized for them.
Here is a chart with several popular GPUs.What is the radeon vs. nvidia performance situation ?
I can't seem to find a good benchmark chart comparing all the current cards on the market.
See the footnotes at the sheet: FahCore_22 performs better than FahCore_21. The sheet still reflects FahCore_21 results.my 1080ti results are better than listed there, I think, but I’m not Sure I get consistent enough work to prove that
Not even sure if folding could honesty push 32 threads?
It can honestly push 128 threads. Maybe more, but I don't have more to try. B-)
I switched Folding@home back on on my GPU-less computers now. I had them running Rosetta, then TN-Grid for a while. Right now at ~10...20% completion I am seeing quite good estimated PPD, much better than on Sunday when I last tried it:Aim for a billion lead as a team goal?
Fri Apr 10 23:00:44 2020
cpu-only1
slot 00 unit 00: RUNNING, 11.26% done, ETA 01:08:00, 930 k ppd
= 930 k ppd
cpu-only2
slot 00 unit 00: RUNNING, 11.26% done, ETA 01:08:00, 931 k ppd
= 931 k ppd
cpu-only3
slot 01 unit 00: RUNNING, 6.10% done, ETA 02:59:00, 237 k ppd
slot 00 unit 01: RUNNING, 10.29% done, ETA 01:11:00, 874 k ppd
= 1.11 M ppd
cpu-only4
slot 00 unit 00: RUNNING, 10.68% done, ETA 01:09:00, 901 k ppd
slot 01 unit 01: RUNNING, 4.33% done, ETA 03:04:00, 233 k ppd
= 1.13 M ppd
cpu-only5
slot 00 unit 00: RUNNING, 23.00% done, ETA 00:25:40, 3.24 M ppd
= 3.24 M ppd
gpu-only1
slot 01 unit 00: RUNNING, 84.70% done, ETA 00:18:37, 1.45 M ppd
slot 00 unit 02: RUNNING, 12.26% done, ETA 02:33:00, 1.46 M ppd
= 2.90 M ppd
gpu-only2
slot 01 unit 00: RUNNING, 12.76% done, ETA 01:43:00, 1.59 M ppd
slot 00 unit 01: RUNNING, 12.42% done, ETA 02:30:00, 1.51 M ppd
= 3.10 M ppd
gpu-only3
slot 01 unit 00: RUNNING, 87.13% done, ETA 00:15:14, 1.47 M ppd
slot 00 unit 02: RUNNING, 2.58% done, ETA 02:45:00, 1.54 M ppd
= 3.01 M ppd
gpu-only4
slot 01 unit 03: RUNNING, 96.08% done, ETA 00:06:59, 1.38 M ppd
slot 00 unit 01: RUNNING, 45.35% done, ETA 01:37:00, 1.43 M ppd
slot 02 unit 04: RUNNING, 60.03% done, ETA 00:51:58, 1.27 M ppd
slot 01 unit 00: READY
= 4.08 M ppd
= 20.44 M ppd, 0 dl, 16 run, 0 ulVS the radeons, it looks like nvidia gets a 25-50% boost with folding compared to some other projects.
Just going by the temp/fan my radeon is not being pushed very hard.
Figures, I got it installed and ready to go and... no work, even with advanced joins setMy 1660 Super arrived today. Will install tonight and then measure at stock for a while (probably the rest of the race with Tom’s). Then I’ll start playing with clock speeds and efficiency.
It's notably lower during the last few hours, but still much better than last Sunday:I switched Folding@home back on on my GPU-less computers now. [...] I am seeing quite good estimated PPD, much better than on Sunday when I last tried it:
56-thread slot: ~0.9 M
88 threads: ~1.1 M combined (partitioned into a 64-thread slot + 24-thread slot)
128-thread slot: ~3.2 M :-O
It's notably lower during the last few hours, but still much better than last Sunday:
56-thread slot: ~0.5...0.9 M
64+24 thread slots: ~0.8...0.9 M combined
128-thread slot: ~1.5...2.6 M
I also had the 128-thread fail just now because a WU could not be scaled as large ("domain decomposition" not compatible), had to reduce the slot to 96 threads temporarily to get it going. This 96-thread slot alone on the 128-thread computer gave 1.0 M PPD.