Anyway, in preparation, I'm trying to figure out what kind of case I might live with. [ . . . ]
I am not sure of the best of the cancer research projects, but the three that I support and use most Often are F@H, then in BOINC, I do Rosetta@home and WCG.I'm allocating $2500 to a cancer-killing machine, planning to build it after the 3990 release for potential increased availability of 3960/3970X parts and hopefully maturity of TRX40 mobos.
What's most important for BOINC-related cancer research DC projects and Folding@Home? Should I care about amount of quad channel memory or should I allocate resources toward the better CPU? What tips have you learned that might help my machine excel?
And @Markfw - do you have resources for ES or otherwise discounted Epyc chips, as I wouldn't be opposed to multiple cheap "moar coar" machines if that's what's important to these DC projects.
My maternal great grandmother had aggressive breast cancer. My maternal grandmother had a rare uterine cancer called MMMT, which killed her within 6 months (passed Dec 21 2018) and there are zero good therapies for it. My maternal uncle had metastatic melanoma refractory to even combination therapies with Keytruda, and he passed last year as well. My own mother has both basal cell skin cancer and stage 1 melanoma. My father in law has metastatic colon cancer in the lungs and brain now. Cancer sucks, and I don't have a way to directly impact cancer research, so rather than donating to Komen (bleh) or another high-overhead organization. I estimate that I can help the ground-level research much more via substantially participating in these DC projects, while also integrating it with a hobby of PC building.I am not sure of the best of the cancer research projects, but the three that I support and use most Often are F@H, then in BOINC, I do Rosetta@home and WCG.
As far as discounted chips, I got burned with 7551 ES chips, as I could not find a motherboard to support them. The 7601 chips I just got are now $900 each, almost the price of discounted 2990wx chips. I have used ES Intel chips a lot, but they are far less powerful and power hungry than the EPYC chips. At the moment, I am not sure which to tell you to get, but the 3900x's are a little bit available, and quite good.
Curious as to what you got excited about cancer research. As you know, I had cancer and it cost me a lot, including my hearing. All I can say is I will support you and give you the best advice I can.
Good to hear. I have a hard time justifying 2990WX given it's last-generation tech, but at the same time I know it would be fantastic. I have some decisions to make!The 3970x would be WAY more powerful than the 2990wx. 2 3950x's would be about the same, but they lack the 4 channel memory (not a big deal). So price and performance are tied, best, 3970x (most expensive), 2 3950x's next (hard to find), and the least expensive and works fairly well is the 2990wx.
Well of course I am a member of TeamAnandtech ! Rosetta I do all projects. WCG I do the caner ones. I am number one for WCG, and getting close to number one on Rosetta.Good to hear. I have a hard time justifying 2990WX given it's last-generation tech, but at the same time I know it would be fantastic. I have some decisions to make!
Are you a member of any groups for F@H (assuming you're using your GPU for it), or for Rosetta/WCG? What project are you contributing to on WCG? Smashing Child Cancer? Cancer Markers?
Think it would be fun to compete on a leaderboard to make it more competitive
I have not checked lately, but it simulates proteins folding and mis-folding (the cause of all cancer)What research is it doing exactly?
A deeper dive:What research is it doing exactly?
Do you know my history ? That I had bladder cancer, they gave me interferon, it did not kill the cancer, and a few months later, virtually overnight I became deaf ? If you did, yes, this is exactly why I spend the money I do on this research. Join us and tell the DC community what we are trying to do. Screw finding big prime numbers. And screw finding aliens. Let cure cancer first.A deeper dive:
There are several categories of things that need research. Just a couple: They look at which cancers express which receptors and how those receptors might change in shape depending on mutations that happen. Similarly, mutated enzyme shapes affect what the cancer cells can and can't do. They look at how proteins might react to interactions with certain types of drugs / immunotherapies. They can also "create" drugs and test them en masse against certain receptors to see which biological shapes/conformations of proteins would interact with, say, a CD20 receptor, so that they can more specifically target with fewer side effects. The amount of work going into it is incredible.
For example, just for one type of lymphoma (DLBCL) there are dozens of known possible mutated enzymes, and knowing how these fold and what they look like is important in order to simulate how they have their effects and how they might be targeted with new therapies. There are specific drugs being created to target a certain type of DLBCL that relies fairly heavily on the type of work F@H and Rosetta are doing. The issue right now is that R-CHOP chemo is kind of effective, but has a ton of side effects. The end goal of this is to find a specific targeted drug with fewer side effects (something I think @Markfw can attest is a huge issue) and be more effective against the cancer itself.
AGREE my man. Watching family die of cancer, agree.Do you know my history ? That I had bladder cancer, they gave me interferon, it did not kill the cancer, and a few months later, virtually overnight I became deaf ? If you did, yes, this is exactly why I spend the money I do on this research. Join us and tell the DC community what we are trying to do. Screw finding big prime numbers. And screw finding aliens. Let cure cancer first.
Here is the URL to our stats. Based on the URL, I think you are correct !AGREE my man. Watching family die of cancer, agree.
For WCG is it the group ID 16700 (TeAm Anandtech) or the other Anandtech one?
Nope, it's all the right direction I think.Here is the URL to our stats. Based on the URL, I think you are correct !
Project Team Stats for TeAm AnandTech in World Community Grid
stats.free-dc.org
And yes, having you family die of cancer can be a great motivator. If you feel I am moving in the wrong direction, please let me know, your research is awesome !
As far as I know, only F@H supports video cards. You think nvidia is supported by WCG ?Nope, it's all the right direction I think.
Joined both Rosetta and WCG cancer projects on my main PC. However, unfortunately on a RX5700 no WCG usage of GPU. May need to drop some $$ on an nVidia card for my cancer killing machine.
Edit:
Some odd spikes in CPU usage...
View attachment 13907
Not sure - but I would rather be able to do both at the same time (WCG on an nvidia card, plus Rosetta on CPU) than having a GPU just sitting there doing nothing.As far as I know, only F@H supports video cards. You think nvidia is supported by WCG ?
That was it! 100% all night.Hi @amrnuke , I believe the above pattern is caused by BOINC being set to use less than 100% of the CPU time. Using Advanced View, look in Options>Computing Preferences, and see if the percentages are set to 100%.
Well, I do F@H on the video card and WCG on the CPU (or Rosetta on the CPU)Not sure - but I would rather be able to do both at the same time (WCG on an nvidia card, plus Rosetta on CPU) than having a GPU just sitting there doing nothing.
That was it! 100% all night.
I'm going to move over to the DC forum since I've veered away from TR for now
Ahhhh I see. Definitely didn't do that the right way. I for some reason was thinking WCG would go on the GPU and totally forgot about F@H. Looks like I'll be setting up F@H tonight on the RX5700. Let's see how it does!Well, I do F@H on the video card and WCG on the CPU (or Rosetta on the CPU)
Cores first.I will do some digging regarding Rosetta@home, WCG, and F@H and see what people have found is most important - cores, clocks, memory bandwidth, amount of memory, etc.
WCG does not have GPU applications at this time, and no known plans for one.unfortunately on a RX5700 no WCG usage of GPU.
If you haven't found the related thread already: It doesn't work yet for F@h, unfortunately.Looks like I'll be setting up F@H tonight on the RX5700. Let's see how it does!