Fast GPU, not enough CPU power to keep up?

[Peter_Hucker]

Since they hide you from criminal activities, and don't log anything, I don't see how they could. To detect you're spamming, they'd have to be watching you.

I was using Ivacy.

[Joe_H]

The VPN providers say they do not "keep" logs, your assumption that it means not logging is not necessarily correct. They do not hide you from criminal activities, just prevent others from knowing your home IP address and ISP. They regularly delete logs, but they know your source IP, what you are transmitting as far as email addresses and everything else, and they know your destination address. The better VPN services will take complaints of spam coming from their supplied IP addresses and backtrack it to drop that customer. That usually shows up on the spam tracking sites as a few uses of that address followed by no further spam reports. The lax ones do not care.

P.S. Ivacy is owned by Gaditek in Karachi, Pakistan. They also admit to owning PureVPN, places that track VPN ownership believe they have two others as well. They also own a number of the VPN review sites such as vpnranks.com, bestvpnservice.com, bestvpn.co, and at least two others.

[Peter_Hucker]

I can tell you with 100% certainty they do hide you from criminal activities. They do not give logs to the police. And no I won't tell you who was doing what.

[calxalot]

They cannot hide that the visible IP address comes from their VPN.

[Peter_Hucker]

Indeed, but if they cannot trace it to you, there is no problem.

As for banning VPN IPs, I did a search and couldn't find any list of known IPs of VPNs, if I were to want to block them from my server.

[muziqaz]

I can tell you with 100% certainty they do hide you from criminal activities. They do not give logs to the police. And no I won't tell you who was doing what.

Now now, unless you own that VPN and run its day to day activities, you cannot state with certainty how companies running VPNs behave

[Peter_Hucker]

Now now, unless you own that VPN and run its day to day activities, you cannot state with certainty how companies running VPNs behave

I have contacts and know the police did not get info they could have used to get someone in big trouble.

[Joe_H]

Indeed, but if they cannot trace it to you, there is no problem.

As for banning VPN IPs, I did a search and couldn't find any list of known IPs of VPNs, if I were to want to block them from my server.

Just because you couldn't find one does not mean they do not exist. There are already content providers who block VPN connections to avoid sending content to countries where it is either not licensed or is prohibited.

[Peter_Hucker]

Just because you couldn't find one does not mean they do not exist. There are already content providers who block VPN connections to avoid sending content to countries where it is either not licensed or is prohibited.

I've never had a problem getting stuff which is banned in my country through a VPN. I guess blocking that country's IPs is enough to please the law. They want to tick the boxes, not actually stop you.

[Alex_Atkin]

v8 does support more than one CPU core on a GPU job, although I'm not sure if this is how/why its going to be used and its not implemented in any cores yet AFAIK.

Some of the features in v8 are there to support future folding core options. Both the OpenMM and GROMACS code the cores are based on have support in recent versions for using both a GPU and CPU together to work on a WU. However testing is needed to see how best to optimize using both resources, and it may depend on the specific parameters connected to various projects. From what I have heard it has been doable on a standalone system where specific hardware can be targeted, but will be harder to implement for systems where that is not known ahead of time.

I'm curious how this will work as I've consistently found that doing any amount of CPU Folding drags down the processing speed of the GPU job.

Its particularly aggressive on my 5950X where I will lose more PPD off the GPU job than I gain from the CPU job, and it doesn't appear to be boost clock related or down to how many cores I use. I can limit CPU Folding to one CCD, the GPU thread will be on the other CCD, but it still happens.

I don't think it has much impact on my 7800X3D system though, so I wonder if its memory latency related and the huge cache negates it?

[muziqaz]

V8 discussion is a pure guesswork, as client is in "experimental" stage with basic functionality working.
PPD in fah is not measure of performance or science throughout.
Yes, you GPU will gain few more points if CPU is sitting idle, however, science will benefit more if you sacrifice those few extra points on GPU slot in order to contribute researchers who have CPU projects available.
When I fold on CPU and GPU I only look how much I lose in terms of TPF, which is always miniscule, at least on AMD GPU side (it might change with HIP core release). Also from contribution point of view, it is wasteful to have 16 core CPU sit idle, and fold just on GPU (before anyone overanalysis this statement: I'm talking from perspective where user comes in with attitude to contribute, because their system is sitting idle anyway, might as well fold, I'm definitely not criticising people who chose to not fold on CPUs).
And according to dev, V8 was initially meant to be automatically taking all your resources and seamlessly using all of them with minimal user input.

[Peter_Hucker]

When I run the CPU on Folding (or another project), I don't see GPU Folding slow down unless I overdo it. Leaving a core or three free is sufficient. The easiest way is to watch the temperature and/or usage of the GPU in something like MSI Afterburner (works on non-MSI cards, and also monitors the CPU).

[Joe_H]

I'm curious how this will work as I've consistently found that doing any amount of CPU Folding drags down the processing speed of the GPU job.

That remains to be seen and really needs further testing. As it stands now, if they use the options some calculations would be offloaded from either the CPU or GPU to the other device. Hopefully that more efficiently uses both. In tests though it has so far been not easy to balance the usage.

[Alex_Atkin]

When I run the CPU on Folding (or another project), I don't see GPU Folding slow down unless I overdo it. Leaving a core or three free is sufficient. The easiest way is to watch the temperature and/or usage of the GPU in something like MSI Afterburner (works on non-MSI cards, and also monitors the CPU).

All I can say is I try it periodically and every time it increases the time it takes the GPU to complete the WU, if the CPU is folding too.

Even having lar.systems stats page open in Chrome drags the speed down.

I'd assume its something wrong with my PC configuration, if it didn't happen on five completely different PCs. But it does seem to impact the 5950X the most, I even tried fiddling with ProcessLasso to try to make sure things are isolated onto different chiplets which I think very slightly helped but not by a lot.

PPD in fah is not measure of performance or science throughout.
Yes, you GPU will gain few more points if CPU is sitting idle, however, science will benefit more if you sacrifice those few extra points on GPU slot in order to contribute researchers who have CPU projects available.

I think its fair to consider PPD a measure of performance, as why else does it exist if its not weighted against which projects are more urgent or more complex? The points system exists entirely to give an element of competitiveness to contributing our energy, so there has to be a logic to how they are dishing out those points.

But that's the thing, were effectively contributing money here, so I'm not going to spend as much on a CPU job as a GPU job, if its less useful. If they are of equal importance then the points should reflect that, as its the only measure we have.

I get that CPU jobs are still important (as in if nobody ran them it would be a huge detriment to certain kinds of research), but is it important enough to slow down the GPU jobs to do them, given the much higher cost per WU in real-cash money. I don't have money to burn, I only fold when it contributes to heating the house and a CPU job at the same time would reach that point quicker, leading to being able to do less GPU jobs. So having some idea of the actual real-world value of what I'm doing would help, but I only have the WU credit to base that on.

[muziqaz]

I have various systems with 3900x, 5950x, 5800x3d and 7950x with various dual or single AMD GPUs. 99% of the projects see a second or 2 increase when CPUs are folding as well. There are occasional projects (once in a blue moon), which do get affected by CPU activity (not as extremely as you describe).
So if you leave 1c or 2 threads to a GPU, you should not see any increased TPF most of the time, unless you have misconfigured your systems and they are running at some weird memory settings or something else is happening to them.
Now PPD: formula works that way that it's value blows up the faster system folds, however you cannot measure exact performance from it.
Sometimes even 10s shorted TPF will result in 2-3m extra PPD.
If you want to compare performance, TPF is the way to go.
CPU work is equally valuable, as it can do all the calculations, while GPU has some limitations however it is much faster.
CPU is your daily driver, GPU is F1 car.
Both of them are vehicles, yet you cannot daily drive F1 car, while your daily driver does everything you need, but slower than F1 car