Uzume

Yes. 4 concurrent work units per card seems to be the sweet spot for both the Titan V and the Vega 64's.Testing suggests Kuroneko and Chibineko will be able to spit out 2.5 PPS-sieve work units per minute between them, not counting CPU chump change. At 24 hours before PrimeGrid start time I will stop other events as the CPU only projects like to cannibalise CPU capacity from GPU work units and stop them running. Coupla hints in the spoiler below, but you guys will likely know this stuff already. app_config.xml

Wow - the first BOINC project this whole contest not run out of a shoebox in the bottom of a closet somewhere in Elbonia.

Sorry I mixed you up with Ithanual with his TR's. Apologies.

Be careful buying outside the TR4's motherboard's memory QVL list for 8x DIMMs unless you know someone else got the same hardware combo working. The list of 8x options on my MB's QVL list was tiny compared to the 4x offerings, and I got bitten by buying off-list (only being able to install 4x of the 8x Corsair 3000 DIMMs I brought).

Nah - its a compat issue. Error details: CUDA RC12!!!!!!!!!! CUDA Device number: 0 CUDA Device: TITAN V Compute capability: 7.0 Multiprocessors: 80 Unsupported CC detected (CC2.0 and better supported only). Each nVidia cuda card has a compute capability number (expand the Cuda Enabled GeForce Products area) and the Titan V is the first with CC7.0, ie, the 10 series are all CC6.1 cards. Asteroids@Home doesn't realise CC7.0 exists yet and gives the "compute capability 2.0 or better supported only" message. I'd post this on their forum but I can't post there without some unspecified number of project credit, and who knows if I will ever reach that unspecified threshhold trying to limp along on CPU work units. Catch 22. Edit: I forgot I have an old crapbox with a 1050ti in it. Will see if I can work up enough credit to post the issue on their forums.

Asteroids@Home + and Titan ▼ = immediate computation errors in work units. No attempt to crunch. This is not unusual in BOINC projects unfortunately, the client doesn't know what to do with them and they are still obscure so there are not enough out in the wild yet to make it an issue. And talk of obscure. Nothing like Googling for an answer to a problem and your own solutionless internet droppings get plastered down the first page. I use the "▼" here so other users also Googling for answers don't get directed to this very post in future.

Definitely if you have the patience. Expect a "1180ti" to have the same cuda power as a Titan V without any of the tensor cores that will never be touched in gaming. I don't see why nVidia would release a weaker 1180ti than it has shown it can already make in the Titan V, unless they plan on further market segmentation with future gaming focused Titans (and I wouldn't put a gaming Titan vs AI Titan marketing mess past them). Especially seeing as 4K monitors need so much more GPU horsepower, a castrated 1180ti makes no sense. Titan V is the best you can get right now, especially if you have a large library of games many of which won't use SLI. Sinking money into dual 1080ti/Titan XPs for only 3 to 4 months gaming seems a waste for any buyer, loaded or not. At least a Titan V will still be good moving forward, BUT, like I said you will wince at having paid 3K for a Titan V when next gen is likely to give you the same performance at less than half the cost in a few months. Be wary of advice of "SLI 1180". Many are forgetting that SLI/Crossfire is being depreciated in favour of mGPU. We don't know yet that SLI will even exist in next gen, yet we do know the entire burden of mGPU implementation is on the game developers, so it is expected to have less utilisation than SLI/Crossfire has had. We have to wait and see.

Done.

Man, every event so far being CPU only is a PITA. Squeezed a bit and now 5 cores on the 1700X and 15 on the 1950X on Universe. Once we get a GPU event I will stop everything else and rack up some points.

I have two systems which use about 1300w running 24/7 compute (SETI) in my room and have just finished the Australian summer. There are two main issues; Noise - you get used to this. But if you share your room with a significant other asking them to get used to it is a very tall order. Also - some people just get bugged no end by background noise, if this is you then also a no go. Heat - not so bad actually as long as you are getting a breeze through your room. So you need a wide open window with another window/door open somewhere in the house to get a crossflow. As soon as you get still conditions then things quickly become intolerable. If you don't get reliable breezes in your room then I would not proceed. Air-con? Chances are you are already going to be losing money mining at this stage in the game with the proliferation of ASICs in the alt-coins which were previously best to GPU mine. Effectively doubling your kilowatt usage by using air con to counter the heat you generate will send you into the red even quicker. Of course if you live in Outer Siberia and could use the heating then this doesn't apply... and good luck moving the cat out of the room.

BREAKTHROUGH LISTEN BEGINS SURVEY OF THE PLANE OF THE MILKY WAY AT PARKES TLDR; Parkes is a CSIRO radio telescope in New South Wales, Australia. It's primary significance is that it is a southern hemisphere receiver, so combined with the northern hemisphere receivers you get full sky coverage. They have been using Parkes data in the past but only from a single beam receiver, this has been upgraded to a 13 beam receiver to vastly broaden the listening capacity. In 2018 the Parkes observations alone are projected to generate 100 petabytes of raw data in 2018 for Seti@Home users to crunch through. Now is a great time for Aussie (and other!) BOINC users to get (back) on board the Seti@Home project... well, maybe after the pentathalon contest is finished anyhow

Titan V's don't SLI but games can theoretically still use more than one of them via mGPU. In practice Civ 6 is the only one I have seen so far that actually does it. If your just gaming without a budget then you can't go past a single Titan V right now for all round performance whether or not a game supports SLI (and how well). It will still be current when the next gen of nVidia consumer cards are released but the retrospective price differential at that time might make you wince. Only consider multiple Titan V if you have a compute addiction like BOINC happening.

That is set by the gpu_usage tag in the app_config.xml file. The very bottom of this URL has the instructions on using the app_config settings. With Seti@Home, I run multiple units through the Titans but the V64's tend to occasionally corrupt results when processing multiple work units at once, so I do them one per card. Might be a project specific issue though.

Set aside 12 cores from Kuroneko and a measly 3 cores from Chibineko (breakdown in sig). Got a few other things on the go as well, but still better than a poke in the eye with a sharp stick.

nvm.

Yah, 24/7 compute apps will quickly show you if there are any cooling issues happening. And it is high temp over time that does damage, not the actual heavy/constant workload itself. The 1700X box I have uses a Corsair H115 and it works well when serviceable, something must have brucked.

Just added U@H project to BOINC and joined team LTT, and set the SETI project to not grab new work units for a while. It isn't like I am in the SETI top 10 or anything, so I can drop rankings for a while and pick them up later.

The University of Queensland (UQ): just a small team but I have a long history with it: https://boincstats.com/en/stats/0/team/detail/144756 Send me a PM with a link to your host stats and I will see if I can see your user sats if you like. Or just post it here if you don't mind a public reply. Also - BOINCstats updates once a day and not always at the same time. Any changes you make won't kick in till tomorrow.

Sorry, playing the 365 day a year contest of BOINCstats rankings and don't want to lose ground

I tried Folding on a mixed Geforce/Radeon (Titan V and Vega 64 LC) configuration and it would not use the Radeon at all. Boinc handles it fine, but Folding doesn't seem to want to play ball when mixing Green and Red teams on a single box. Caveat: would need a larger sample size of mixed systems to confirm this, but it isn't that common for people to mix GPU teams like this.

CPU contests are just distributing bunkering organisation contests. It is only with all the GPUs churning through the entire queue multiple times a day that you start getting a fairly decent contest.

Your being deliberately obtuse now. The entire point of GPP is to give Nvidia exclusive access to gaming brands by making the AIB's an offer they can't refuse. Yes, AIBs can ignore it and be sent to Nvidia chip procurement purgatory.

That is my understanding of the word "exclusively" but I am not a smart man.

Reading that was what gave me that impression in the first place. Specifically the stipulation of: "Gaming Brand Aligned Exclusively With GeForce."