Posts by robsmith

21) Message boards : GPUs : PCI express risers to use multiple GPUs on one motherboard - not detecting card? (Message 95606)
Posted 25 days ago by robsmith
Post:
Not really, it depends on what part of the GPU GPU-Z is using to determine the percentage use. For most GPUs it is possible to max out one part while other parts are a long way off fully loaded. Further, each part of a GPU will have its own, internal, power budget and that all those added together can exceed the total power that the internal PSU system is capable of supplying reliably.
22) Message boards : Questions and problems : Projects whose tasks take less time to execute (Message 95595)
Posted 25 days ago by robsmith
Post:
The answer depends on your computer hardware and operating system. For example a SETI@Home task will take less than minute a Linux system with a RTX2080 GPU while the same task will take over 8 hours on an old mobile phone.
23) Message boards : GPUs : PCI express risers to use multiple GPUs on one motherboard - not detecting card? (Message 95297)
Posted 18 Jan 2020 by robsmith
Post:
Well, there are no publishable pictures of the complete beast. Indeed the complete beast is very boring to look at, just a large box with power, data and cooling connections.
Initially a small system was tested using RTX2080 to give an idea of what feeders were going to be needed. Next tests were with earlier Quadro which left the RTX2080 behind, after six months (and some mods to the cooling) the RTX8000 were installed, and they are a step up again. The trouble with bench marks and specs is they don't always reflect what happens in real life under very high stress.

Being a totally air-cooled system the GPUs were obtained without their fans, etc. blast air at ~4C keeps everything in check.

But we digress.
24) Message boards : GPUs : PCI express risers to use multiple GPUs on one motherboard - not detecting card? (Message 95294)
Posted 18 Jan 2020 by robsmith
Post:
Picking up from what Richard said - the SETI CUDA applications have a very efficient task slicing and management which reduces the demands on the bus. However the SETI SoG application is very much more demanding on the bus, in part this is down to the technology used (CUDA vs.OpenCL). So when considering the performance required of the bus one has to be aware of what the application's needs are.

If I&S thinks he's seen some high GPU count computers - how about 256 Quadro rtx 800s in one computer, each one is sitting on its own x16 PCI link, to keep that lot fed and under control there are a whole host of Xeons, ARMs and FPGAs, not to mention the cooling system that shifts over 100kW of heat away (and no doubt warms a fair bit of the rest of the building....)
25) Message boards : Questions and problems : ARM based 8 core design, use all 8 cores possible? (Message 95259)
Posted 18 Jan 2020 by robsmith
Post:
That's going to be highly dependent on the types of core - it appears that those on the device being demonstrated are all the same, which is unlike many phones & pads. But it is always worth a try.
26) Message boards : The Lounge : Grumbles and Glory (Message 95231)
Posted 17 Jan 2020 by robsmith
Post:
...about six feet ;-)
Seriously it varies vastly from person to person, anything from hours to months. Just take it easy, I hope you are at the "few days" end of the spectrum.
27) Message boards : The Lounge : Help Desk Expert? (Message 95223)
Posted 17 Jan 2020 by robsmith
Post:
I've not reached that figure yet, but then I do delete loads.
28) Message boards : Questions and problems : Why doesn't Boinc schedule earlier deadlines first? (Message 95187)
Posted 16 Jan 2020 by robsmith
Post:
What you are talking about is the "panic" mode - where a task that is in danger of not being completed in time gets promoted up the queue - and this was already mentioned in one of my posts last night.
There is a known downside to this - if a project sends out tasks with gross under-estimates of its "FLOP count" then it's guessed run-time will be too short, couple that with a very close "return by" date while you have other tasks from other projects with reasonably accurate expected run-times BOINC can get very confused - particularly if another project does much the same thing at the same time. This is very much an issue with the project(s) concerned, they should be more realistic in both setting the target dates and the FLOPS count.
Then there is the situation where a project sends out work with impossible run-time and return-date combinations - sending a task that will take 3 days, but a deadline in two days. No amount of queue jumping will find that extra day :-(
29) Message boards : GPUs : PCI express risers to use multiple GPUs on one motherboard - not detecting card? (Message 95182)
Posted 16 Jan 2020 by robsmith
Post:
There are several configurations of "two ribbon" cable risers.
Those where each side of the connector is connected to its own cable. These do give a bit of signal segregation, so may be less noisy than just a single cable, but if not done properly could cause timing errors.
Those which double-up the power lines. Good idea to have a bit more copper available for the power
Those which use both cables for all signals. No real advantage over just doubling-up on the power line, but may cause some signal timing errors.

Given 6 inches is about a quarter-wavelength at the frequencies these buses run at the use of unscreened cables may be OK on some motherboards, but may be a disaster on others. It is always best to use screened cables. (Also the screened cables I've seen have tended to have been of a far better manufacturing quality than unscreened cables.)
30) Message boards : Questions and problems : Why doesn't Boinc schedule earlier deadlines first? (Message 95163)
Posted 15 Jan 2020 by robsmith
Post:
That is not what was said or implied by prodigit's post about torrents -- he said they change priority during operation, which is correct for some torrent tools, and that BOINC should do the same. Actually that is a bit of a simplistic view in that the performance of a large torrent download appears to increase as it gets nearer completion, and a real change as the segment location and place routine has less work to do as the torrent gets towards the end (fewer empty slots to fill). Either of these may appear to be a change in priority.
This is very different to the way BOINC operates. In "normal" mode there is a fixed sequence of tasks, first-in-first-out which applies to both task-swap and task end. There is a "panic" mode, which is that if BOINC notices that a that isn't running is probably not going to finish by its deadline it is promoted up the queue and gets run at either the next available task swap, or when a task completes.
31) Message boards : GPUs : PCI express risers to use multiple GPUs on one motherboard - not detecting card? (Message 95160)
Posted 15 Jan 2020 by robsmith
Post:
What about if I (as I do already to a certain extent), run several tasks on each GPU? Do you know what the continuous average data rate is for GPU projects? Running several tasks on one GPU would get rid of problems with peak transfer rate, just as at the moment I use it to get around the slow CPU.


That all depends on a whole pile of factors - some of which are: The GPU in question; The application(s) involved; The number of tasks/applications being run; The motherboard; The CPU; The motherboard RAM. I have no doubt that others can add to the list.

Let's go a bit deeper into the list.
The GPU - obviously the more powerful the GPU the more likely it is that it will be able to run multiple tasks
The application - a bit less obvious, but some applications have been designed to use all of one of the GPU's resources, thus trying to run two of them will cause quite a substantial amount of bus traffic. There are undoubtedly pairs of applications that are just so incompatible with each other that they will simply refuse to play ball.
The motherboard - fairly obvious, if this has a very poor PCIE bus, or another data bottleneck in the path between the CPU and the GPU then loading it more won't exactly help the overall performance.
The CPU - simply, some CPUs are better at supporting the demands of GPUs than others - for example the AMD FX family is pretty bad when compared with its contemporary Intel equivalents.
Motherboard RAM - is there enough of it - particularly important where the GPU shares RAM space with the CPU.

In truth the only way one finds out is try it and see - start with a low multiple and don't be surprised at the result (in either direction). Allowing plenty of time (days or even weeks) for a stable situation to be reached, and obviously be prepared to abandon any combinations or multiples that cause lots of errors to be generated.

Your second point sounds very much like a hardware/BIOS issue with your motherboard, some combinations will actually allow you to see a display from both GPUs during boot, others won't. When you've got two GPUs installed one way to find out if it is a slot dependency is to have a single monitor and see what happens during boot, trying both GPUs in turn, you may find that either GPU can be used for boot purposes, or only one of them. Weirdest one I had was a system with three GPUs where I could use any of them as the "boot GPU", but when I took one of the GPUs only the one is the slot nearest the CPU could be used As you can imagine it was "quite interesting" when the two GPUs were in the slots further away.
32) Message boards : Questions and problems : Why doesn't Boinc schedule earlier deadlines first? (Message 95156)
Posted 15 Jan 2020 by robsmith
Post:
And the fact that what was implied, or said, is totally and utterly wrong?
That shows that you do not understand what BOINC is about, and what it can and cannot do, or what it should, or should not do.

If you want to change BOINC I suggest that you join the BOINC steering group, properly explain to them your ideas, and listen to their explanation, and don't be surprised if you are grilled on your ideas. Some may join the very long list of possible enhancements, and some may be rejected, such is the way of things....
33) Message boards : Documentation : documentation needs an update (Message 95140)
Posted 15 Jan 2020 by robsmith
Post:
But not very accurate - very few modern desktops draw LESS than 50w at "idle", more typically its 75-100W.
Laptops are certainly getting down that way (if not lower), and SMALL GPU servers can be running up to 2-3kW even with "modern" GPUs.
Recently I did the thermal management system for a very large GPU server, running a large number of RTX8000 processors, and that is sitting at between 95-125kW (Ouch!!! - I'm glad I'm not paying that power bill).

Certainly that page does need updating, so if you feel inclined, and can get some accurate figures, rather than "assumptions" then I'm sure the BOINC documentation team would be all to happy to hear from you.
34) Message boards : Questions and problems : Is anyone using windows 98 for a project? (Message 95138)
Posted 15 Jan 2020 by robsmith
Post:
Yet again you ignore the OP's quest - I suggest you READ what the OP wants to achieve before suggstign anything else.
35) Message boards : GPUs : PCI express risers to use multiple GPUs on one motherboard - not detecting card? (Message 95137)
Posted 15 Jan 2020 by robsmith
Post:
Great, but if you read the description of what the OP's motherboard has, and what he want to achieve, then you would realise that what you are suggesting will not fully answer his quest.
Summary - The motherboard has two x16 slots and a number of x1 slots. He wants to run more than 2 GPUs, he wants to have them air-cooled, and the x16 slots are too close together for his comfort. Thus he needs to be able to use the at least one of the x16 slots with a riser, and the x1 slots will need a riser anyway - and those riser will have to be x1 to x16 because they have to be x1 at the motherboard end. So far he's had no joy in getting the motherboard to recognise anything sitting on a x1 to x16 riser. Ian&Steve has made a few suggestions, and I know he has done a lot of work in getting similar (not identical) systems working.
36) Message boards : Questions and problems : Why doesn't Boinc schedule earlier deadlines first? (Message 95136)
Posted 15 Jan 2020 by robsmith
Post:
And as such has nothing to do with BOINC, but might have a lot to do with each and every project that uses computers to do calculations of any sort.
Remember BOINC does none of the science work, it only provides an environment in which some science may be done.


Agreed. Therefore it's on topic.


Well, you've talked yourself into the self-contradictory stance - as it has nothing to do with BOINC then it should NOT be being discussed here, but on the PROJECT forum.
37) Message boards : Questions and problems : Why doesn't Boinc schedule earlier deadlines first? (Message 95134)
Posted 15 Jan 2020 by robsmith
Post:
Stop changing your mind about what is being talked about - YOU raised the subject of Torrent priorities, and when someone responds explaining why one cannot compare Torrent priorities with the priorities used by other types of application YOU object.
38) Message boards : The Lounge : The Seti is Down Cafe (Message 95106)
Posted 15 Jan 2020 by robsmith
Post:
Who was it said "four days" - they have a lot to answer for.
39) Message boards : Questions and problems : Why doesn't Boinc schedule earlier deadlines first? (Message 95055)
Posted 14 Jan 2020 by robsmith
Post:
And as such has nothing to do with BOINC, but might have a lot to do with each and every project that uses computers to do calculations of any sort.
Remember BOINC does none of the science work, it only provides an environment in which some science may be done.
40) Message boards : GPUs : PCI express risers to use multiple GPUs on one motherboard - not detecting card? (Message 95033)
Posted 14 Jan 2020 by robsmith
Post:
Why do I have to plug them into 16x slots if they are only 1x plugs?

You don't - electrically the first 18(?) pins on an x16 are exactly the same as those on an x1 - or at least SHOULD be. The reason for trying in an x16 is to make sure that there isn't something strange with the x1 sockets (and that wouldn't be the first time)


Previous 20 · Next 20

Copyright © 2020 University of California. Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.