Running DiffusionBee on my M1 Mac mini while looking at PowerMetric in another window, I'm getting about 9.5 watts of CPU+GPU power usage for about thirty seconds per image generated, so 0.00008 kilowatt-hours.
154 publicly visible posts • joined 7 Jun 2008
Re: GPT-4 is a thing
But the GPT-3 paper described the number of parameters and the compute intensity of the training, whilst the GPT-4 paper decided to be deliberately uselessly vague about that to free up pages to fill with useless analysis of 'AI risk' and of how they had crippled the model so that it didn't regurgitate bomb- or drug-making instructions which could be found in moments with an obvious Google search.
If the technology is more efficient, you can stick more of it in a box to use the same amount of power at the same temperature. There was a brief period where people used individual low-power servers, before realising that collecting jobs using virtual machines onto big machines hosting 256 vCPU in 2U and 1kW was a much more efficient use of the very-expensive space.
This has been a serious problem in Eastern Europe where district heating was provided by coal-fired power stations or by steelworks uneconomic in a global context. On the whole if people stop wanting to host computers in London we have some more serious problems, and replacing servers with electric resistance heaters is an ugly but effective solution. A one-kilowatt resistor costs about £50 compared to a £50,000 Sapphire-Rapids-plus-H100 server.
Re: Two years to tender
Not really - what's bankrupted Birmingham is spending years underpaying their female employees and then having to find £700 million upfront to pay the second part of the settlement.
£100 million on an Oracle migration that was budgeted at £19 million and didn't work is a drop in the bucket compared to that.
Is CXL pricing really as keen as claimed here?
I needed a lot of memory in a server last year, so bought 12x16GB DDR4 sticks from bargainhardware for £42 each. The going rate is now £42 per stick for 32GB sticks, if I wanted 1536GB it would be a thousand pounds even if I threw away all the current DDR4.
It seems unlikely that a PCIe card with a controller significantly more complicated than the normal kind and 256GB of brand new memory chips can compete with that on pricing.
Is this a weird artificial market brought into existence by Dell's insane pricing on DDR5?
It's a pity that universities do not seem temperamentally suited to 'we had a vast and wide-ranging disaster with our Oracle implementation; we have managed to make it work reasonably well now; we will spin-off the better people in our internal IT group as a consultancy to help other universities have smaller and more confined disasters'.
Yes, UK universities are each a weird thing unto themselves, but they resemble one another rather more than any of them resembles the strongly hierarchically structured Wisconsin widget-works which SAP or Oracle ERP start off expecting to model.
A particular CAPSA problem seemed to be that 'a person capable of signing off expenses' and 'a thing against which expenses could be signed off' were both very heavyweight structures because there were expected to be about six of them in the company, whilst in a university every researcher got to be capable of signing off expenses against their own separate grant.
You don't have to. But releasing a marginal improvement annually still means that people replacing their 2018 smartphone with a 2024 smartphone get a nicer experience than the people who replaced their 2015 smartphone with a 2021 smartphone, and so on; as long as any chip company proceeds to the next ARM core, their competitors will be obliged either to do the same or to lower their prices. You can still get a brand new Samsung phone with eight A55 cores.
The chips have been out with live hyperscale customers for a year, end-users willing to sign NDAs have had access to them in some of the clouds for three months, they have gone through twelve separate steppings, Intel has taken a half-billion-dollar charge on unshippable product (and is currently rearranging itself internally so that the processor group doesn't get to run new steppings through the fab at its own convenience); I think it is fair to call them battle-hardened now.
I'd always thought of 'edge' as 'can we run this chunky calculation on the nice fast ARM core that the user has already paid for in their smartphone, rather than on a no-better ARM core that AWS is charging us four cents an hour for' - if you're procuring new hardware for edge then you're doing it wrong.
Re: Batteries not included
And Artemis was a particularly bad setup for cubesats, because they were installed in July 2021 and then there was no access to them, even for battery recharging, until the launch sixteen months later. 80% functionality rate for things stuck in a drawer for sixteen months is not bad ...
It's a perfectly normal high-quality overpriced beige coat from the front (https://us.burberry.com/monogram-motif-waterloo-trench-coat-p80647751) but with a really surprisingly ugly white creeper-face design in the small of the back.
I was thinking Burberry might be charging that much for a Burberry-check in-game skin, which would have been a bit more than averagely silly at the height of the NFT nonsense and absolutely ridiculous now.
Re: For those unfamiliar with Qualcomm lawyers..
In particular, Qualcomm are accusing Arm of wanting to move to a per-device licensing model *like Qualcomm's*.
The 'no Arm extra IP blocks without an Arm CPU' part doesn't seem completely unreasonable, though I think Intel did do an x86-with-Mali SoC at one point - if Arm are willing to miss out on Mali revenue because they are worried about RISCV+Mali SoCs that's up to them, The claim that they might be moving to 'no non-Arm IP blocks on a chip with an Arm CPU' is obviously complete nonsense since Arm don't make memory or PCIe controllers.
That seems a tiny machine to be boasting about
0.4 petaflops, so a fifth the compute of the smallest machine in the June 2022 top500, and with 196 processors and no GPUs I would be startled if it took more than one rack (Frontier packs 128 processors and 512 GPUs per rack). I am surprised HPE bothered putting out a press release!
There are three larger supercomputers announced on the top500 in Singapore already.
In what form does this PCIe 6.0 Interface Subsystem come? Any PHY with fast SERDES is basically analogue design and very deeply process-specific at the moment, it would be nice to know what processes the blocks are available for.
(the press release at Rambus just says 'on advanced process nodes', it would be an interesting insight into the fabrication industry to know whether that includes Intel Integrated Foundry Services and Samsung's offerings, or just means TSMC N5 and N3)
Oh no, an extra year before fanciest-available datacentre CPUs trickle down through the second-hand market to add to the medium-performance computing facility in my garage.
On the other hand, an enforced one-year gap on computer acquisition just as electricity bills are tripling is probably not the worst plan.
Re: Girding of Loins
They've already done it - Amazon has warehouses full of its c7g Arm-based units, Apple sells Arm processors by the million.
Or is the only interpretation of "taking on Intel head-on" that you'd accept one in which Arm itself sells physical objects in the retail market to plug into sockets on motherboards, which they've been absolutely clear for twenty years they're never going to do.
Re: Mixer services
They're really not, because keeping cryptocurrency transactions private is not a legitimate goal.
The weird lend-real-money-against-cryptocurrency industry already regards 'coins that have gone through a mixer' as unacceptable collateral; I would expect that in the medium term it will not be possible to sell for real money coins that have ever passed through a mixer.
At the price charged for Teslas, and the number of GPUs they have in them anyway, wouldn't a second touchscreen in the front mounted somewhere that the passenger can see it but the driver not be a sensible approach? It would also allow the driver's touchscreen to be moved somewhere that the driver isn't having to look away from the centreline to see it.
Re: A Kludge
The whole point of hyper threading is that it works well on code which *hasn't* been thoroughly optimised. If you've got two vector instructions lined up for each tick, hyper threading can't get you anything; if your thread is waiting two hundred ticks for the L3 cache to divulge the next operand, having another thread running until it too needs to wait for the L3 cache is extremely helpful.
I am very sad at the way that people are using fairly hypothetical security arguments to disable the features that make processors actually good at computing: I am willing to do my banking on my phone if that means my actual computers can crunch numbers at a higher percentage of its peak speed.
I was, accidentally, there for the changeover; all the banks and ATMs were closed for a couple of days, which was somewhat irritating since I'd turned up without any cash because Romanian lei are hard to come by.
Particularly irritating was that, when the ATMs reopened, they were still dispensing the old notes!
Putting on my pedant hat, ESA had been launching the Galileo satellites on Soyuz rockets owned by Arianespace and launched from their site in French Guyana; so Roscosmos got the money for the launch vehicle to pass through to the manufacturer, but was not responsible for the launch.
(the two earliest Giove satellites did go up from Baikonur)
It's not completely obvious that launching 675kg satellites four at a time on Ariane 5 (which can happily take six tons to the harder-to-reach geosynchronous orbit) is better than launching two at a time on a smaller rocket.
A competing manufacturer of annoying squishy expensive pavements says '5W continuous', so I think that particular problem is just a matter of a badly-written press release.
The problem is that a 5W generator produces £5.70 of electricity a year at retail price, and it seems wildly unlikely that the maintenance of the annoying squishy pavement, let alone the interest on a loan to cover the install price, is of that order of magnitude.
If they're trying to fit a petaflop in five racks for $3 million, the processors will just be sitting there driving the GPUs. My guess is that they're anticipating nVidia, who has quite tight links with POWER, producing a POWER-and-GPU box which might be more cost-effective than the current Xeon-and-GPU box.
($3 million is interesting close to what you get if you divide a petaflop by the current $129k 42.4TF nVidia DGX100 and expect a small discount for buying in bulk)
Re: The problem with es-16-xg is not just the SFP+ cables
This is a comment guaranteed to not fill you with delight, but: have you tried the cable in more than one port on your switch? I have some cables that work only in ports 6 and 8 (and, accordingly, a bag of fibre transceivers winging its way to me from fs.com)
And switch vendors seem to think that 'doesn't work with some brands of DAC in some ports' is a problem they address by maintaining a compatibility list, or in the worse cases by mandating you use cables provided by them and made out of mark-up coated in a thick layer of dielectric expense, rather than by fixing their rassenfrassen switch hardware.
Mine's the one with the distress-purchase fibre transceivers in the pocket
Re: hate DACs
Sadly the cheap server motherboards have integrated 10G with SFP+ connectors, and SFP+ to 10GBaseT transceivers don't exist because driving 10GBaseT requires more power than SFP+ is specced to provide.
If you're buying the network card, 10GBaseT is not a bad choice (if you're going to have the exotic cabling faff, at least go for 40Gbit Infiniband and have the extra speed); if it's integrated then you're screwed.
Re: ARM is not very popular as desktop
Since Conroe I have expected PCs to last forever. Builds have grown big enough that it's nice on a software-development workstation to have 2GB per thread and ideal to have 4GB per thread, but that's been possible since Haswell.
A problem is that memory controller IP, which is very black magic and comes from a very small number of suppliers, charges a big premium for each extra address line; so it would be a significant extra expense to make a phone or set-top-box SoC which happens to be able to support 16GB, and therefore you're not going to be able to make a nice 16GB devboard around a ubiquitous SoC.
Re: Let's step back for a second
So you have to stop offering really crappy free WiFi. Where's the issue? If your free WiFi is not making your coffee shop a fiver a day in custom, why are you bothering? When almost everyone in the coffee shop will be accessing your wifi using a mobile device which can happily switch to 3G, providing a service worse than 3G is a waste of time.
Re: It's already all wrong....
And Gaia watches them move and measures how far they've moved, so two million of the stars are annotated with velocities already and about a hundred million will be by the end of the mission.
The researchers did do the obvious experiment to see whether any velocities were changing over time, because that would be an exciting result, but didn't find any convincing examples - it turned out simply to be an excellent way to find one category of mistakes matching up stars in the fifteen-year-old Tycho catalogue.
Hipparcos data analysis is a really hairy problem - van Leeuwen spent a decade inventing new data-reduction techniques and re-running the whole analysis, the fact that the satellite was stuck in geostationary transfer orbit because its apogee motor failed to fire didn't help.
Gaia's estimate for the position of the Pleiades is in one of the free-access papers published today, and is unsurprisingly absolutely bang in the middle of the results from Earth-bound telescopes or from astrophysical arguments.
The error in this data release is 0.0002 arc-seconds and the intended error at the end of mission is 0.000007 arc-seconds. Star positions to an accuracy of 7 arc-seconds can be achieved with a DSLR and a reasonable telephoto lens, no need for a two-billion-Euro mission at the L2 Lagrange point.
I'm afraid you've screwed up the BCE->CE correction
It seems you have changed 'BCE' to 'CE' throughout. The Assyrian eclipse was in 763 BCE, we are reasonably confident that the foundation of Rome was 753 BCE, the first year of the Gonghe Regency, after which Chinese chronology is apparently pretty well-known, is 841BCE.
This means the last three paragraphs of the article, which I suspect are commentary inserted by the article author rather than extracted from the source publication, don't make much sense.
You're 1500 years out!
The Miyake event was in 775 CE (during the reign of Offa, king of Mercia, and at the end of the reign of Constantine V, Emperor of Byzantium), not 775BCE (around the time of the first Olympic Games in classical Greece, and well before the invention of either Mercia or Byzantium). The other one is 993-994CE (reign of Ethelred the Unready)
There seems to be some Enterprisey over-engineering going on here; I rather like the idea of a card with a PCIe switch and four or six M.2 NVMe slots on it, so I can run software RAID0 across six cheap Samsung half-terabyte NVMe cards. IDT make a switch chip that would be perfect for this, which costs about $200, so I'd expect a competent Guangzhou shanzhai to make a profit selling the card for £250.