Problems and Technical Issues with Rosetta@home

Message boards : Number crunching : Problems and Technical Issues with Rosetta@home

To post messages, you must log in.

Previous · 1 . . . 166 · 167 · 168 · 169 · 170 · 171 · 172 . . . 219 · Next

AuthorMessage
Sid Celery

Send message
Joined: 11 Feb 08
Posts: 1832
Credit: 33,790,392
RAC: 7,545
Message 104509 - Posted: 26 Jan 2022, 16:28:19 UTC - in response to Message 104335.  

Just checking in because I had a fair few Rosetta 4.20 tasks come down.
But I think they already ran out...

I'm useful like that
YOU!! You stole them! I wanted those. I'm going to hunt you down, and I mean physically!

I actually did. Full buffer on both machines I have near me before mentioning it.
No need to thank me.
I'll make tea - do you take sugar?
I don't like hot drinks. Orange juice or vodka please, or both.

I ran into some trouble with vodka many years ago, so I don't keep it (or anything) around any more. Looks like it's orange for you.

Which I'll need to get as I managed to grab a whole lot more Rosetta tasks earlier today - until they ran out again...
ID: 104509 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sid Celery

Send message
Joined: 11 Feb 08
Posts: 1832
Credit: 33,790,392
RAC: 7,545
Message 104510 - Posted: 26 Jan 2022, 16:37:20 UTC - in response to Message 104357.  

Robetta, as far as I can tell, is separate from Rosetta@home and is used mostly by researchers outside of the Baker Lab/IPD. It's an interface for users who wish to get computing power for their jobs.
Jobs that require the use of Rosetta 4.20 that are submitted to Robetta get sent to Rosetta@home but the rest goes to the other servers that they set up when they launched RoseTTAFold.

While that may be largely true, it may not be exclusively true.
If you see "bcov" in the task name I believe that's Brian Coventry who is (or certainly was at one time) a researcher at Rosetta.
He posted a few messages here when the Covid research started in March & April 2020 and was clearly right in the thick of development here
ID: 104510 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sid Celery

Send message
Joined: 11 Feb 08
Posts: 1832
Credit: 33,790,392
RAC: 7,545
Message 104511 - Posted: 26 Jan 2022, 16:42:30 UTC - in response to Message 104376.  

Sid Celery posted something a few months ago that he received from Admin or someone like that who said that the Python job that had been submitted by one of the IPD researchers was "huge".
It's not that big, it's only a few million tasks. I've seen the queue at 15 million. But maybe that was several projects at once.

That was true. I think they meant 2.2m was <relatively> huge for Python tasks, because it's also true we've had 20m Rosetta tasks to wade through from around April last year when the RAM demands started going through the roof.
Which kind of makes Peter's point again
ID: 104511 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Falconet

Send message
Joined: 9 Mar 09
Posts: 342
Credit: 987,274
RAC: 0
Message 104512 - Posted: 26 Jan 2022, 17:25:06 UTC - in response to Message 104510.  
Last modified: 26 Jan 2022, 17:32:04 UTC

Robetta, as far as I can tell, is separate from Rosetta@home and is used mostly by researchers outside of the Baker Lab/IPD. It's an interface for users who wish to get computing power for their jobs.
Jobs that require the use of Rosetta 4.20 that are submitted to Robetta get sent to Rosetta@home but the rest goes to the other servers that they set up when they launched RoseTTAFold.

While that may be largely true, it may not be exclusively true.
If you see "bcov" in the task name I believe that's Brian Coventry who is (or certainly was at one time) a researcher at Rosetta.
He posted a few messages here when the Covid research started in March & April 2020 and was clearly right in the thick of development here



Yes, that's why I said "mostly". I've seen those "bcov" tasks many times but I don't think they were submitted via Robetta since they had no rb_ name.
Edit: Searching "bcov" as a user name on Robetta doesn't yield any results.

I do remember that the first Coronavirus work units came from Robetta, having been submitted by something called the Seattle Infectious Disease Research Center or something like that and it was related/joint work to Baker Lab/IPD or at least UW. If I recall correctly, of course.


Edit: It's called the Seattle Structural Genomics Center for Infectious Disease (SSGCID). Seems like it's part of a Consortium including UW and there was joint work with the Baker Lab per this page and this was one of the first COVID jobs .

Things moved fast back then.
ID: 104512 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Peter Hucker of the Scottish Boinc Team
Avatar

Send message
Joined: 12 Aug 06
Posts: 1183
Credit: 5,645,510
RAC: 1,825
Message 104514 - Posted: 26 Jan 2022, 18:54:28 UTC - in response to Message 104502.  

Looking at the book, it appears I have 6 SATA ports, so if I lose 1 to the m.2 I still have 1 left.

That's ok.
What do you think?
You will not lose a SATA port. Your M2 drive is not SATA. M2 has two connections in it, PCI-Express and SATA. The slot on your board will take two different types of drive. Only the SATA drives eat a SATA port. Yours is PCI-Express so eats PCI-Express lanes. On the second M2 socket, it will disable the third PCI-Express socket, which you aren't using, so that's fine. All your SATA ports will continue to function as normal. Plug your M2 drive into the second socket and leave everything else as it is. This should allow everything to run normally.
ID: 104514 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Peter Hucker of the Scottish Boinc Team
Avatar

Send message
Joined: 12 Aug 06
Posts: 1183
Credit: 5,645,510
RAC: 1,825
Message 104515 - Posted: 26 Jan 2022, 18:56:06 UTC - in response to Message 104509.  

I don't like hot drinks. Orange juice or vodka please, or both.
I ran into some trouble with vodka many years ago, so I don't keep it (or anything) around any more. Looks like it's orange for you.
What trouble did Vodka cause you?

Which I'll need to get as I managed to grab a whole lot more Rosetta tasks earlier today - until they ran out again...
I read that as Ribena, I must be thirsty.
ID: 104515 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Greg_BE
Avatar

Send message
Joined: 30 May 06
Posts: 5502
Credit: 5,468,187
RAC: 1,837
Message 104516 - Posted: 26 Jan 2022, 20:30:05 UTC - in response to Message 104514.  

Looking at the book, it appears I have 6 SATA ports, so if I lose 1 to the m.2 I still have 1 left.

That's ok.
What do you think?
You will not lose a SATA port. Your M2 drive is not SATA. M2 has two connections in it, PCI-Express and SATA. The slot on your board will take two different types of drive. Only the SATA drives eat a SATA port. Yours is PCI-Express so eats PCI-Express lanes. On the second M2 socket, it will disable the third PCI-Express socket, which you aren't using, so that's fine. All your SATA ports will continue to function as normal. Plug your M2 drive into the second socket and leave everything else as it is. This should allow everything to run normally.



Nope. Because the PCIe lanes are full on the CPU.

But one other question, where does the PCI slots board on that expansion unit go?
I see the little one that goes in the slot, but the big one? (the 4x PCI-USB)
ID: 104516 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Peter Hucker of the Scottish Boinc Team
Avatar

Send message
Joined: 12 Aug 06
Posts: 1183
Credit: 5,645,510
RAC: 1,825
Message 104517 - Posted: 26 Jan 2022, 20:35:43 UTC - in response to Message 104516.  

Looking at the book, it appears I have 6 SATA ports, so if I lose 1 to the m.2 I still have 1 left.

That's ok.
What do you think?
You will not lose a SATA port. Your M2 drive is not SATA. M2 has two connections in it, PCI-Express and SATA. The slot on your board will take two different types of drive. Only the SATA drives eat a SATA port. Yours is PCI-Express so eats PCI-Express lanes. On the second M2 socket, it will disable the third PCI-Express socket, which you aren't using, so that's fine. All your SATA ports will continue to function as normal. Plug your M2 drive into the second socket and leave everything else as it is. This should allow everything to run normally.



Nope. Because the PCIe lanes are full on the CPU.

But one other question, where does the PCI slots board on that expansion unit go?
I see the little one that goes in the slot, but the big one? (the 4x PCI-USB)
They are not all full. You have 24. 4 to the chipset. 8 to GPU 1. 8 to GPU 2. 4 to either the second M2 slot or the third GPU socket. You are not using those 4. Lanes cannot just go wherever you want them, they are only wired to go to certain places. You're missing out on 4 of them. Just move it to the second one.

The "big one" is just 4 individual ones sat side by side, the same as the big one in my other link to the 1x PCI-USB. They go anywhere you like they're just connectors for the GPU to sit in.
ID: 104517 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Greg_BE
Avatar

Send message
Joined: 30 May 06
Posts: 5502
Credit: 5,468,187
RAC: 1,837
Message 104519 - Posted: 26 Jan 2022, 20:56:03 UTC - in response to Message 104517.  
Last modified: 26 Jan 2022, 20:57:30 UTC

Looking at the book, it appears I have 6 SATA ports, so if I lose 1 to the m.2 I still have 1 left.

That's ok.
What do you think?
You will not lose a SATA port. Your M2 drive is not SATA. M2 has two connections in it, PCI-Express and SATA. The slot on your board will take two different types of drive. Only the SATA drives eat a SATA port. Yours is PCI-Express so eats PCI-Express lanes. On the second M2 socket, it will disable the third PCI-Express socket, which you aren't using, so that's fine. All your SATA ports will continue to function as normal. Plug your M2 drive into the second socket and leave everything else as it is. This should allow everything to run normally.



Nope. Because the PCIe lanes are full on the CPU.

But one other question, where does the PCI slots board on that expansion unit go?
I see the little one that goes in the slot, but the big one? (the 4x PCI-USB)
They are not all full. You have 24. 4 to the chipset. 8 to GPU 1. 8 to GPU 2. 4 to either the second M2 slot or the third GPU socket. You are not using those 4. Lanes cannot just go wherever you want them, they are only wired to go to certain places. You're missing out on 4 of them. Just move it to the second one.

The "big one" is just 4 individual ones sat side by side, the same as the big one in my other link to the 1x PCI-USB. They go anywhere you like they're just connectors for the GPU to sit in.



M.2 #2 same problem.
If 16 channels are in use for GPU and 4 for CPU where are the other 4?

And I still get: (m.2 #1 and #2)
1/26/2022 9:05:53 PM | | GPU detection failed: process exited with status 0x1: Incorrect function. (0x1)
1/26/2022 9:05:53 PM | | No usable GPUs found

As for your extension kit, It looks like I can get 2 singles for around 25 Euro from Germany and I don't have to pay import fees like the ones from China.
ID: 104519 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Peter Hucker of the Scottish Boinc Team
Avatar

Send message
Joined: 12 Aug 06
Posts: 1183
Credit: 5,645,510
RAC: 1,825
Message 104520 - Posted: 26 Jan 2022, 20:56:08 UTC - in response to Message 104517.  
Last modified: 26 Jan 2022, 21:00:19 UTC

Ah, you've already tried it in the second slot. Something weird is going on here. Like you, I see 4 lanes missing.

Ok, so what happens if you remove one GPU? Does the M2 drive and the other GPU work in Boinc then?

Also test the firmware update again with a GPU out, this is also pointing to something up with it. It's a pity you don't have another machine you can test it in. Friend or colleague willing?

I wonder..... incorrect function. Could Boinc be stupid enough to think your drive is a GPU because of it being on PCI express? So it fails to talk to it like a GPU, and doesn't bother checking for more of them?
ID: 104520 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Peter Hucker of the Scottish Boinc Team
Avatar

Send message
Joined: 12 Aug 06
Posts: 1183
Credit: 5,645,510
RAC: 1,825
Message 104521 - Posted: 26 Jan 2022, 21:22:16 UTC - in response to Message 104519.  

Luckily the UK doesn't have import fees on low priced items, I think only stuff over £25 gets them, and I can usually persuade the sender to lie about the value or write "sample" or "returned faulty goods". I got some of mine from China, but then found a bitcoin (or some kind of coin) guy in the UK selling off a load of them, I got a pack of 10 which turned out to be 12 when he found some more for a fraction of the new price.

Please try only 1 GPU and the M2 drive first, to make sure nothing else is going on. There could be a faulty drive, which you need to send back soon to avoid hassle? If that setup works ok, then it's lanes, although we're both at a loss to where the missing 4 lanes went. If it's lanes, then the adapters will be good, it'll drop the GPUs from using 8 each to 1 each, then everything should work. You just need to work out where to mount them.
ID: 104521 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Greg_BE
Avatar

Send message
Joined: 30 May 06
Posts: 5502
Credit: 5,468,187
RAC: 1,837
Message 104522 - Posted: 26 Jan 2022, 22:22:36 UTC - in response to Message 104521.  

Luckily the UK doesn't have import fees on low priced items, I think only stuff over £25 gets them, and I can usually persuade the sender to lie about the value or write "sample" or "returned faulty goods". I got some of mine from China, but then found a bitcoin (or some kind of coin) guy in the UK selling off a load of them, I got a pack of 10 which turned out to be 12 when he found some more for a fraction of the new price.

Please try only 1 GPU and the M2 drive first, to make sure nothing else is going on. There could be a faulty drive, which you need to send back soon to avoid hassle? If that setup works ok, then it's lanes, although we're both at a loss to where the missing 4 lanes went. If it's lanes, then the adapters will be good, it'll drop the GPUs from using 8 each to 1 each, then everything should work. You just need to work out where to mount them.


It's getting messy. And your the only one with real information. The tech just gave me the same old information that I already knew. Can't really get Samsung involved just yet. I guess it's AMD I will have to ask now. But I'll try the one GPU tomorrow night.

But I don't think changing location of one GPU to the third PCIe 16 slot will make a difference.

As for import fees, remember, Brexit separated you from EU rules. Stuff purchased from the UK outside of the big players (Amazon etc) are also subject to tax. EU member to EU member purchases are excluded. It used to be the way you described, but then Allibaba dumping pissed of the EU and they passed these new rules.
ID: 104522 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Peter Hucker of the Scottish Boinc Team
Avatar

Send message
Joined: 12 Aug 06
Posts: 1183
Credit: 5,645,510
RAC: 1,825
Message 104523 - Posted: 26 Jan 2022, 22:38:52 UTC - in response to Message 104522.  

If it works with one GPU and the drive, and the other GPU not connected, then it's lanes. Then you know where you stand - you could put one GPU on an extension to create 7 free lanes.

I didn't suggest moving a GPU to the third slot, because you'd said earlier that was physically difficult. But if you can do that, the third slot only has 4 lanes (it's called a 16 slot due to it's size, but if you shine a torch into it you'll see the connectors only go part way across), so you'd reduce the GPU to 4 lanes to give 4 to the drive. That could work, but you'd have to put the drive in the first M2 slot, as the second M2 slot shares lanes with the 3rd GPU slot.

I'm so glad I left the EU. Way too many rules and complications. I sold something to an EU guy on Ebay. I forget which country it was, but he paid three taxes on it, totalling more than the product! Import tax, sales tax, internet tax, something like that.
ID: 104523 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Greg_BE
Avatar

Send message
Joined: 30 May 06
Posts: 5502
Credit: 5,468,187
RAC: 1,837
Message 104525 - Posted: 26 Jan 2022, 23:27:49 UTC - in response to Message 104523.  

If it works with one GPU and the drive, and the other GPU not connected, then it's lanes. Then you know where you stand - you could put one GPU on an extension to create 7 free lanes.

I didn't suggest moving a GPU to the third slot, because you'd said earlier that was physically difficult. But if you can do that, the third slot only has 4 lanes (it's called a 16 slot due to it's size, but if you shine a torch into it you'll see the connectors only go part way across), so you'd reduce the GPU to 4 lanes to give 4 to the drive. That could work, but you'd have to put the drive in the first M2 slot, as the second M2 slot shares lanes with the 3rd GPU slot.

I'm so glad I left the EU. Way too many rules and complications. I sold something to an EU guy on Ebay. I forget which country it was, but he paid three taxes on it, totalling more than the product! Import tax, sales tax, internet tax, something like that.



Internet tax? That could be the anti alibaba tax thing.
Official documentation for Belgium just says 15 fees and 21% on items value (sometimes I think it is a perceived value or minimum tax because they "estimated" the value to be around 30 euros for something they didn't even look at. Trust me I gave them hell about that.
ID: 104525 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Peter Hucker of the Scottish Boinc Team
Avatar

Send message
Joined: 12 Aug 06
Posts: 1183
Credit: 5,645,510
RAC: 1,825
Message 104526 - Posted: 26 Jan 2022, 23:47:12 UTC - in response to Message 104525.  

Internet tax? That could be the anti alibaba tax thing.
Official documentation for Belgium just says 15 fees and 21% on items value (sometimes I think it is a perceived value or minimum tax because they "estimated" the value to be around 30 euros for something they didn't even look at. Trust me I gave them hell about that.
I did once get taxed heavily for something. I bought a cheap item (about £20) from China. By way of an apology as they were out of stock, they first sent me a gift. It was a pointless 99p necklace or badge or something, but the idiot seller put "jewellery, value £40" on the customs label. I was out at the time so it got taken to the local post depot. I assumed it was the product, and thought I'd get the money back from the seller, so I went and paid the tax to the post office and collected it. I was also charged an admin fee by the post office. When I found this stupid little thing that cost me 100 times what it's worth, I went mental at the seller. The real product then arrived the next day with a value of £2 written on it. I photographed both things and the packaging and he paid up. It cost him a fortune.
ID: 104526 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Greg_BE
Avatar

Send message
Joined: 30 May 06
Posts: 5502
Credit: 5,468,187
RAC: 1,837
Message 104528 - Posted: 27 Jan 2022, 7:04:56 UTC - in response to Message 104526.  

Internet tax? That could be the anti alibaba tax thing.
Official documentation for Belgium just says 15 fees and 21% on items value (sometimes I think it is a perceived value or minimum tax because they "estimated" the value to be around 30 euros for something they didn't even look at. Trust me I gave them hell about that.
I did once get taxed heavily for something. I bought a cheap item (about £20) from China. By way of an apology as they were out of stock, they first sent me a gift. It was a pointless 99p necklace or badge or something, but the idiot seller put "jewellery, value £40" on the customs label. I was out at the time so it got taken to the local post depot. I assumed it was the product, and thought I'd get the money back from the seller, so I went and paid the tax to the post office and collected it. I was also charged an admin fee by the post office. When I found this stupid little thing that cost me 100 times what it's worth, I went mental at the seller. The real product then arrived the next day with a value of £2 written on it. I photographed both things and the packaging and he paid up. It cost him a fortune.



I have never bought from China direct.
Takes to long. That expander would take a month to get here.
And the savings were not that great.
Oh well....
I'll change the GPU's tonight. Last try.
ID: 104528 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Greg_BE
Avatar

Send message
Joined: 30 May 06
Posts: 5502
Credit: 5,468,187
RAC: 1,837
Message 104532 - Posted: 27 Jan 2022, 18:20:43 UTC

Well taking the 1050 out did not solve the problem.
So I guess it goes to my original theory that the drive is defective.
Going to finish off the last work and take the drive out and see what happens on reinstall.
ID: 104532 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Peter Hucker of the Scottish Boinc Team
Avatar

Send message
Joined: 12 Aug 06
Posts: 1183
Credit: 5,645,510
RAC: 1,825
Message 104533 - Posted: 27 Jan 2022, 18:26:46 UTC - in response to Message 104528.  

I have never bought from China direct.
Takes to long. That expander would take a month to get here.
I just use them when I'm not in a hurry, like stocking up on something. Or stuff I'll need for a project later on.
ID: 104533 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Peter Hucker of the Scottish Boinc Team
Avatar

Send message
Joined: 12 Aug 06
Posts: 1183
Credit: 5,645,510
RAC: 1,825
Message 104534 - Posted: 27 Jan 2022, 18:29:56 UTC - in response to Message 104532.  
Last modified: 27 Jan 2022, 18:32:02 UTC

Well taking the 1050 out did not solve the problem.
So I guess it goes to my original theory that the drive is defective.
Going to finish off the last work and take the drive out and see what happens on reinstall.
Did you have the drive in M2 slot 2 when you tried a single GPU? Otherwise I'm thinking it might be trying to take the GPU's lanes.

Actually I think it won't make any difference removing a GPU. Two GPUs use 2x8 lanes. One GPU uses 1x16 lanes. There is no way your board and CPU are designed so they can't cope with one GPU and one M2 drive. If you have the drive in slot 2 and it fails, it's busted. Add in you not being able to flash it. Something funny going on with that drive. I got a Crucial SSD drive once (just on SATA) and it suddenly failed to be recognised. "Known fault, flash it" they said. I couldn't flash what wasn't detected. Back it went.
ID: 104534 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Greg_BE
Avatar

Send message
Joined: 30 May 06
Posts: 5502
Credit: 5,468,187
RAC: 1,837
Message 104535 - Posted: 27 Jan 2022, 20:02:07 UTC - in response to Message 104534.  

Well taking the 1050 out did not solve the problem.
So I guess it goes to my original theory that the drive is defective.
Going to finish off the last work and take the drive out and see what happens on reinstall.
Did you have the drive in M2 slot 2 when you tried a single GPU? Otherwise I'm thinking it might be trying to take the GPU's lanes.

Actually I think it won't make any difference removing a GPU. Two GPUs use 2x8 lanes. One GPU uses 1x16 lanes. There is no way your board and CPU are designed so they can't cope with one GPU and one M2 drive. If you have the drive in slot 2 and it fails, it's busted. Add in you not being able to flash it. Something funny going on with that drive. I got a Crucial SSD drive once (just on SATA) and it suddenly failed to be recognised. "Known fault, flash it" they said. I couldn't flash what wasn't detected. Back it went.



I'm done with all this hardware swap nonsense. Going to box it up when things are done and swap it out.
ID: 104535 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Previous · 1 . . . 166 · 167 · 168 · 169 · 170 · 171 · 172 . . . 219 · Next

Message boards : Number crunching : Problems and Technical Issues with Rosetta@home



©2022 University of Washington
https://www.bakerlab.org