An intel 12900k (Gen 12) compared to a 2600k (Gen 2, launched 2011) is about 120% faster or a bit over 2 times in single threaded applications, those +5-15% uplifts every generation add up over time but its nothing like the earlier years when they might double in performance in a single generation.
It really depends if that application uses AES 256 bit and other modern instructions. The 12900k has 16 cores vs 4 of the 2600k, although 8 of those extra cores are E-cores. This performance increase doesn't necessarily come from free given the application may need to be adjusted to utilise those extra cores especially when half of them are slower to ensure the workload is distributed properly.
Even within a vertical scaling by getting a new processor for just single threaded applications its interesting that much of the big benefits come from targeting the new instructions and then the new cores. Both of which may require source updates to get significant performance uplift from.
> This performance increase doesn't necessarily come from free given the application may need to be adjusted to utilise those extra cores especially when half of them are slower to ensure the workload is distributed properly.
It especially doesn't come for free when you consider that 12900k uses nearly 2.5x the power of a 2600k at peak.
I'm not even sure 12900k can operate at full load under air cooling for longer than a few minutes.
> is about 120% faster or a bit over 2 times in single threaded applications
1. Doesn't that also account for speedups in memory and I/O?
2. Even if the app is single-threaded, the OS isn't, so unless it's very very inactive other than the foreground application (which is possible), there might still be an effect of the higher core count.
Funnily enough, most apps aren't taking enough advantage of multi-core multi-threading environments that are common across all major platforms.
The single biggest bottleneck to improvement is the general lack of developers using the APIs to the fullest extent when designing applications. Its not really hardware anymore.
Though, to the points being made, we aren't seeing the 18 month doubling like we did in the earlier decades of computing.
Unless you're multitasking, the OS on a separate thread gets you about 5-10% speedup. It's not really noteworthy.
Unless you lived through the 1990s I don't think you understand how fast things were improving. Routine doubling of scores every 18 months is an insane thing. In 1990 the state of the art was 8mhz chips. By 2002, the state of the art was a 5ghz chip. So almost a thousand times faster in a decade.
Are chips now a thousand times faster than they were in 2015? No they are not.
When the parent comment says “OS on a separate thread”, they’re talking about the fact that even if your user-space workload is single-threaded, the OS has a separate set of threads, which can run on separate cores from the user thread.
We need a significant rethink in how new materials are released into the world. The current process where a business can just release something new and it goes through little in the way of real safety testing can't continue. As we build ever more exotic materials the potential for catastrophic and rapid danger is high. We have destroyed and disrupted our habitat too much already we need to stop doing this. We seriously need to stop producing microplastic and clean up what we have unleashed on the world quickly.
BYD is slowly gaining but its not much. Tesla had only slightly better sales in january 2024 and its quite variable month to month its not indicative of a collapse in sales yet. Need to see this continue before anyone can say Tesla sales have actually declined rapidly.
In most cases today if we don't attribute a direct crime solely to one person but instead to an organisation everyone avoids criminal prosecutions. Its only the people who didn't manage to spread the blame through the rest of the organisation that go down.
Its really hard to compare SD cards and especially durability because we get no information as to what they are doing differently. You can get a better idea of the performance characteristics than the broad categories (A1 or A2, largely useless) with a review on storagereview but they don't have anything further or a way to compare durability. It matters less in cameras but for Single board computers or dashcam uses it would be nice to have a better idea of the practical durability and the usage pattern that would preserve it.
I had bzip2 eat some files some years back and ended up loosing them. I don't trust bzip2 for backup files and I wouldn't trust bzip3 either as a result.
Most of the time is still in the hops. Every machine along the route and transitions into fibre and electrical cables is what is adding the time. In theory we can do drastically better than we do today with faster routing machines but latency hasn’t really improved very much for long connections despite all the advancements in the past 30 years.
Nvidia is the only one that does this on the sort of scale these AI companies need. They do boxes packed full of AI accelerators with custom high speed bus connectors and high end interconnectivity to make super computers full of GPUs very easy to purchase combined with the software to make it all work. No one else has that sort of integration to the very large that is necessary for the sort of training the top firms are doing.
For inference, Nvidia's strengths seem to be not very important.
You can do inference on a single GPU. And AFAIK the software stack is not important for inference either. Because you don't have to experiment with the sofware. You just need to get it to run and then you will run it for a long time unchanged. Groq for example runs LLAMA on their custom hardware, correct?
And I expect hardware for inference to become a bigger market than hardware for training.
The only reason you need all those GPUs is because they only have a fraction of the ram you can cram in a server.
With AMD focusing on ram channels and cores the above rig can do 6-8 tokens per second inference.
The GPUs will be faster, but the point is inference on the top deepseek model is possible for $6k with an AMD server rig. 8 H200's alone would cost $256,000 and gobble up way more power than the 400 watt envelope of that EPYC rig.
A lot of businesses view human errors that a system has as something to complain at the users about and put policy around to stop it from happening, this does not work. Equally banning an essential technology doesn't solve the problem because people still need a mechanism for sending messages asynchronously to each other.
The issue is about the user interface of the clients. The problem with email is that no one really makes any money out of it. Its a mature and stagnant technology and it gets little investment. People want to pay as little as possible for something that you can get from Google for "free". But the issues are all to do with the user interface and defending people from relatively rare errors because people misclicked and misunderstood the impact.There are very common features of email that should be nice and quick and there are features rarely used that cause a lot of problems and the happy path should be slowed and made more difficult to do accidentally and without knowledge of the potential impact.
But the issue is the maturity and lack of investment because otherwise these problem features like CC would be handled better and there are numerous ways it could be made safer and errors a lot less likely.
You can buy loose tea and get metal tea stirrers that contain it which is about the best way to avoid the plastic. Still comes sealed in a plastic bag though, its very hard to avoid plastics at all.
It really depends if that application uses AES 256 bit and other modern instructions. The 12900k has 16 cores vs 4 of the 2600k, although 8 of those extra cores are E-cores. This performance increase doesn't necessarily come from free given the application may need to be adjusted to utilise those extra cores especially when half of them are slower to ensure the workload is distributed properly.
Even within a vertical scaling by getting a new processor for just single threaded applications its interesting that much of the big benefits come from targeting the new instructions and then the new cores. Both of which may require source updates to get significant performance uplift from.
https://www.cpu-monkey.com/en/compare_cpu-intel_core_i7_1270...
reply