RTX 3080 vx RX 6900XT vs RTX3090

[tulanebarandgrill]

I'd be interested to know who out there is running at about a 20M PPD level and what hardware it takes to do that.

Or frankly anyone > 5M PPD, what are hardware combinations that are known working for 8, 10, 12, 15, or 20M PPD ?

I see some people getting those kinds of numbers and I'm not sure if it is one of these blockchain miner setups or if they just have several computers running like I do.

1. Look for a user with enough PPD, you will find some here: https://folding.extremeoverclocking.com ... ist.php?s=
2. Type in the user name here: https://apps.foldingathome.org/cpu

So I did what you suggested and I found this :

C711D85E66C982CF xxx.xxx.xxx.129 266078 NVidia GPU Linux 155,556 1 2020-11-22 11:24:50 17421 0 726 92
DB11D85E66C982CF xxx.xxx.xxx.129 266078 x86 Linux 24,159 1 2020-11-22 11:54:54 16925 38 7238 21
D111D85E66C982CF xxx.xxx.xxx.129 266078 x86 Linux 9,403.75 1 2020-11-22 12:04:51 17415 0 296 180
C811D85E66C982CF xxx.xxx.xxx.129 266078 NVidia GPU Linux 245,888 1 2020-11-22 12:14:53 13428 12702 35 1
CA11D85E66C982CF xxx.xxx.xxx.129 266078 NVidia GPU Linux 200,951 1 2020-11-22 12:25:00 14908 172 6 170
C911D85E66C982CF xxx.xxx.xxx.129 266078 NVidia GPU Linux 248,010 1 2020-11-22 12:44:53 13428 12699 24 1
7881D55E20AF6CFC xxx.xxx.xxx.129 266078 NVidia GPU Linux 161,116 1 2020-11-27 05:05:01 17425 0 958 33
7A81D55E20AF6CFC xxx.xxx.xxx.129 266078 NVidia GPU Linux 150,282 1 2020-11-27 05:25:02 17421 0 463 84
8181D55E20AF6CFC xxx.xxx.xxx.129 266078 x86 Linux 25,179 1 2020-11-27 05:34:59 16925 22 1302 43
7981D55E20AF6CFC xxx.xxx.xxx.129 266078 NVidia GPU Linux 196,910 1 2020-11-27 05:34:59 14907 491 5 154
8B81D55E20AF6CFC xxx.xxx.xxx.129 266078 x86 Linux 9,097.5 1 2020-11-27 05:55:01 17408 0 946 235
7781D55E20AF6CFC xxx.xxx.xxx.129 266078 NVidia GPU Linux 152,874 1 2020-11-27 05:55:02 17420 0 1515 64'


Am I right in assuming this person has 8 GPUs ? possibly in 2 different systems ?

[bruce]

From information posted elsewhere, AMD distinguishes between models using the same chipset by a revision ID number. F@h only uses the manufacturer and device ID number fields of a PCI device ID in GPUs.txt. So for now there is no way to tell which model is being used.

This may or may not change at some point in the future. There is some discussion on changing how the client detects a GPU for use and that may show up in a future version or update to the client software.

Has this been documented in an enhancement request on github? IT's not going to happen unless we make our wishes known.

It certainly would be helpful in distinguishing between these guys.
Ellesmere XT [Radeon RX 470/480/570/580/590]

[Jupis_folder]

C711D85E66C982CF xxx.xxx.xxx.129 266078 NVidia GPU Linux 155,556 1 2020-11-22 11:24:50 17421 0 726 92
DB11D85E66C982CF xxx.xxx.xxx.129 266078 x86 Linux 24,159 1 2020-11-22 11:54:54 16925 38 7238 21
D111D85E66C982CF xxx.xxx.xxx.129 266078 x86 Linux 9,403.75 1 2020-11-22 12:04:51 17415 0 296 180
C811D85E66C982CF xxx.xxx.xxx.129 266078 NVidia GPU Linux 245,888 1 2020-11-22 12:14:53 13428 12702 35 1
CA11D85E66C982CF xxx.xxx.xxx.129 266078 NVidia GPU Linux 200,951 1 2020-11-22 12:25:00 14908 172 6 170
C911D85E66C982CF xxx.xxx.xxx.129 266078 NVidia GPU Linux 248,010 1 2020-11-22 12:44:53 13428 12699 24 1
7881D55E20AF6CFC xxx.xxx.xxx.129 266078 NVidia GPU Linux 161,116 1 2020-11-27 05:05:01 17425 0 958 33
7A81D55E20AF6CFC xxx.xxx.xxx.129 266078 NVidia GPU Linux 150,282 1 2020-11-27 05:25:02 17421 0 463 84
8181D55E20AF6CFC xxx.xxx.xxx.129 266078 x86 Linux 25,179 1 2020-11-27 05:34:59 16925 22 1302 43
7981D55E20AF6CFC xxx.xxx.xxx.129 266078 NVidia GPU Linux 196,910 1 2020-11-27 05:34:59 14907 491 5 154
8B81D55E20AF6CFC xxx.xxx.xxx.129 266078 x86 Linux 9,097.5 1 2020-11-27 05:55:01 17408 0 946 235
7781D55E20AF6CFC xxx.xxx.xxx.129 266078 NVidia GPU Linux 152,874 1 2020-11-27 05:55:02 17420 0 1515 64'


Am I right in assuming this person has 8 GPUs ? possibly in 2 different systems ?

8 GPUs, each of class RTX2070 or better. And 4 CPU folding slots. If you consider PCIe lanes and power supply, it looks like that user uses 4 systems, each equipped with 2 GPUs.

[Neil-B]

I'd go for two twin cpu (prob xeon) servers with 4x GPUs on each from what I am seeing

[Jupis_folder]

So we have to read the CPUid in this way:
C911-> Folding slot
D85E66C982CF -> System (e.g. mac address)
?

[ajm]

I don't know about the composition of the CPUid, but with this method you have to consider the date (last returned) as well. https://apps.foldingathome.org/cpu lists everything the user folded with [EDIT:]in the last 90 days, even what is now gone.
Here, we see that the 6 first lines refer to WUs returned earlier. Only the 6 last ones were active when you checked the data. Maybe the system was changed, or the first lines show elements that were deactivated or paused after 2020-11-22.

[MeeLee]

I think it's 4 GPUs, and 2 CPUs. Only the results from 11/27 count. The results from 11/22 don't count, as they are way past the deadline. (He could have stopped folding on the 22nd though, but what's more likely is that he did a system upgrade after the 22nd, and started folding again with that same system, now having a different CPU ID.

It's been too long for me to say,but the PC ID is somewhere in this: "81D55E20AF6CFC". It's previous configuration was in this: "11D85E66C982CF"
The slot ID is the first 2 digits, with CPU ID being '8B', and '81'.
Seeing that all GPUs start with a 7, it's possible that we're talking about 1 system here, using 1 type of PCIE slot (directly connected to the CPU), not 2 separate systems or 2 different type of PCIE slots.
Sometimes x16 slots get one starting number (7 in this case), but x1 slots, m.2 slots, or x4 slots, get another starting number.
'77', '79', '78', and '7A' are 4 PCIE slots, presuming (x16) PCIE slot 01, 02, 03, and 04; since 7, 8, 9, and A, are consecutive numbers in Hex, the PCIE slots will most likely also be consecutive.

[bruce]

As far as I've been able to determine, the CPUid is an 8-byte random number which is used for slot 0. To distinguish between work completed by various slots, the slot number is added to the first byte. (second unpacked nibble). I don't believe there's an other hidden meanings imbedded in the number.

[Nathan_P]

I'd go for two twin cpu (prob xeon) servers with 4x GPUs on each from what I am seeing

There's no need for dual xeon's these days, threadripper or epyc can handle 4 gpu's with ease. My next upgrade will see me reitring my dual xeon machine and replacing it with my current daily driver, a 1920x. I've have more GPU x16 lanes, lower power consumption and if I upgrade the cpu to a 2950x better PPD from a cpu slot

[Neil-B]

I'd go for two twin cpu (prob xeon) servers with 4x GPUs on each from what I am seeing

There's no need for dual xeon's these days, threadripper or epyc can handle 4 gpu's with ease. My next upgrade will see me reitring my dual xeon machine and replacing it with my current daily driver, a 1920x. I've have more GPU x16 lanes, lower power consumption and if I upgrade the cpu to a 2950x better PPD from a cpu slot

Think you misunderstood .. I was stating what setup I thought had produced the listing from the cpu app that was posted/being discussed !!

[MeeLee]

Just an update on release dates,
Aside from Covid, memory chip scarceness, and scalper sales people, there's a third factor to account for low availability, and it's not the reviewers samples...
Apparently the RTX3000 and RX6000 series are scavenged by miners, to replace their rtx2000 series gpus.
I did some research, and a 3070 does about 10% higher hash rate than a 2080Ti, all the while running at 25W less power. And a 3080 does about 25% higher hashrate, for close to the same power consumption (little higher).

Just like the RTX 2000 series were scavenged in the west, while China got the GTX1000 series,
Now the west invests in RTX 3000, and China is buying mostly RTX 2000 series GPUs for mining purposes.
They're especially fond of 2060s apparently...

Miners didn't have their fill yet, and gotten most of the pre-holiday releases.
You can still buy them at 2x MSRP on Amazon, but I would estimate prices to stabilize around February the earliest!
Unless you are lucky and can snatch one from a good source, but those usually won't be available for long.