|
I'm hoping the memory shortages continue. Made a 50 grand bet on Micron at the start of the year and it's doubled since then. Pretty crazy
|
# ¿ Nov 3, 2017 09:55 |
|
|
# ¿ May 9, 2024 10:36 |
|
Linus laying a burn on Intel https://lkml.org/lkml/2018/1/3/797 quote:Why is this all done without any configuration options?
|
# ¿ Jan 4, 2018 15:24 |
|
I still have a lot of trouble with audio on Ubuntu but these days with my Bluetooth headset
|
# ¿ Feb 11, 2019 07:42 |
|
There's an entire market of prosumers to small tech outfits that buy parts from places like Amazon and that would absolutely see material performance hits on applications ranging from GCC compilation to video encoding to even stuff like 7zip compression. A benchmark article from last year specifically on Intel SMT https://www.phoronix.com/scan.php?page=article&item=intel-ht-2018&num=1 quote:Long story short, Hyper Threading is still very much relevant in 2018 with current-generation Intel CPUs. In the threaded workloads that could scale past a few threads, HT/SMT on this Core i7 8700K processor yielded about a 30% performance improvement in many of these real-world test cases.
|
# ¿ May 17, 2019 05:08 |
|
It might be the case that the majority of users aren't realistically under threat but there's still uncertainty whether Intel or OS providers (Apple, Google, Microsoft) will force you to take a performance hit to cover their bases. ChromeOS disabling HT by default is a minor example but is Intel going to keep separate branches of its software/firmware updates moving forward for people that don't want to take any performance hit? They've already specifically stated that their soft mitigation causes up to an 8% performance hit.
|
# ¿ May 17, 2019 06:51 |
|
MaxxBot posted:Phoronix has the hard numbers, here's the tldr. lol...but hey, users can disable these mitigations if they want to so no problemo
|
# ¿ May 19, 2019 02:15 |
|
https://twitter.com/atomicthumbs/status/1203771427679166464?s=20
|
# ¿ Dec 13, 2019 08:15 |
|
What's up with Intel? Is it a management problem or did they make a wrong bet on tech sometime back and it's taking time to turn the ship around.
|
# ¿ Jul 24, 2020 10:01 |
|
Err... Intel has pretty good support for ML inferencing (AVX2 and their OpenVINO platform).
|
# ¿ Jul 25, 2020 12:22 |
|
RE ML and CPUs, inferencing is moving to INT8 quantized operations. GPUs will always be better for training but in production when you're serving inferencing queries to end-users, people shift to CPU since they're cheaper and scale up better.
|
# ¿ Jul 25, 2020 23:57 |
|
Yah, obviously it'll vary by use-case. I do NLP stuff - we do training on a mix of local GPUs, cloud TPUs but we've migrated to CPU-based inferencing mostly for scale + cost.
|
# ¿ Jul 26, 2020 10:05 |
|
I've spent a fair bit of time recently on transferring GPU-trained models to either CPUs or embedded platforms with cut-down tensor cores for inferencing. They're tenable for a lot of use-cases. The thrust of research and work in this area is to compress or distill the models in an efficient manner without losing much if any performance. Common techniques include throwing out precision (FP32->FP16->INT8), pruning them (throwing out swathes of weights), or distilling them (using the trained model as a teacher to transfer its knowledge to a smaller model with an efficient architecture). Research has shown that a good way to build ML models is to train large and deep models then compress them down to fit your inferencing platform.
|
# ¿ Jul 27, 2020 01:18 |
|
Intel's a large labyrinthine bureaucracy at this point with 100K employees. It's pretty difficult to transform an organization like that, and mass firings or replacing the top layer with outsiders is unlikely to be the answer. I suspect you could parachute in Jensen Huang and he'd make a mess of things as an outsider without history in the company and a group of mid and higher tier experienced staff he could lean on.
|
# ¿ Jul 29, 2020 00:00 |
|
The market is dumb. They're pricing AMD and Nvidia at tech bubble levels so let's see where we end up.
|
# ¿ Jul 29, 2020 02:12 |
|
it's possible to believe that AMD and Nvidia have good products and also that their stocks are priced at dumb levels. take a look at their p/e ratios. i guess i'm blessed to be an overpaid computer toucher but the US stock market specifically feels like late stage capitalism given everything else going on in the country.
|
# ¿ Jul 30, 2020 01:17 |
|
i'm not going to complain about my stock portfolio but number continually going up isn't going to do much for me if my neighborhood gets burnt down in the eviction riots of 2021.
|
# ¿ Jul 30, 2020 02:12 |
|
i went from a MBA to a Razer Blade to a Razer Stealth back to an MBP. i don't get people who say Windows laptops are better than Macs even at the same price point
|
# ¿ Aug 4, 2020 02:20 |
|
I had the pre-2020 Stealth 13. I was chasing the dream of a mobile gaming laptop for a while but they just suck due to the physics of thermal dissipation. I don't get how people accept the fan noise that a gaming laptop outputs. Even the ROG G14 which is the current sweetheart has the same issue. Went back to MBP + a Switch.
|
# ¿ Aug 4, 2020 02:34 |
|
Have there been any good business write-ups about Intel falling apart? Something more focused on the whys rather than the technical details.
|
# ¿ Aug 9, 2020 08:03 |
|
People aren’t willing to pay for long form text tech reviews so... I’m just happy there is some (any) decent tech coverage on video (DigitalFoundry etc)
|
# ¿ Aug 9, 2020 20:27 |
|
Nice meltdown as they say
|
# ¿ Aug 10, 2020 00:13 |
|
Murthy’s Law
|
# ¿ Aug 10, 2020 06:49 |
|
https://twitter.com/harukaze5719/status/1319643238513270785?s=20 lol, did some bad news just come out for Intel?
|
# ¿ Oct 23, 2020 15:25 |
|
apple approaching their PCs from a mobile first perspective kinda limits where they'll go. they don't seem interested in cloud compute and i'm skeptical they're interested in making a serious play for the HEDT/local server space. they can make a play for any consumer electronics/computer space they care about so it's more a matter of management focus and entering a market which is either strategically important or actually pushes their financials. the m1's strength is its battery life - otoh, it's a pity all that processing power is limited to their software ecosystem. outside of a limited set of creative software, you can't really apply the power on stuff like gaming.
|
# ¿ Nov 23, 2020 01:30 |
|
Laslow posted:Yep. Cloud is a race to the bottom and they don’t play that game. lol cloud is a major profit center for big tech companies extending to even hardware makers that cater to cloud providers
|
# ¿ Nov 23, 2020 06:13 |
|
Ampere's a side show - intel/amd/nvidia should be more concerned about the major cloud players rolling their own silicon e.g. Amazon/graviton, Google/TPUs. Intel and Nvidia making juicy margins on cloud hardware was never a long-term thing. at some point, the cloud service providers are going to pull an Apple.
|
# ¿ Dec 24, 2020 04:56 |
|
movax posted:Google doesn't have any adults in the room to capitalize on their resources / advantages / custom silicon, I feel. lol, do you have a macro replacing "cloud" with "my butt"?
|
# ¿ Dec 24, 2020 06:27 |
|
i was thinking more in terms of AI. they've got a research complex pushing not just papers but hardware/architectural improvements. i get the impression their researchers get ready access to TPU pods (1024 TPUv3s) which has been an advantage in churning out benchmark beating models. a hardware improvement that i've personally encountered is google/TPUs switching to their own half-precision floating point type (bfloat16). a lot of their research models are trained on it these days and you often can't run them on nvidia hardware without switching to fp32. nvidia's only introducing it with Ampere cards and as far as i can tell, the major ML framework don't support it on their cards yet.
|
# ¿ Dec 24, 2020 06:51 |
|
Intel's been a backwater for engineering talent for a while now. Even a decade ago when I was in school, I remember the semicon companies pulling in the middling graduates. The comp gap's only ballooned since then.
|
# ¿ Dec 31, 2020 00:48 |
|
zero use for AVX-512 in gaming given GPUs even for inferencing
|
# ¿ Jan 3, 2021 02:12 |
|
i'm referring specifically to AVX-512. also, does it still throttle performance on mixed workloads?
|
# ¿ Jan 3, 2021 03:23 |
|
Paul MaudDib posted:not on Ice Lake at least. quote:Licence-based downclocking is only one source of downclocking. It is also possible to hit power, thermal or current limits. Some configurations may only be able to run wide SIMD instructions on all cores for a short period of time before exceeding running power limits.
|
# ¿ Jan 3, 2021 04:02 |
|
lol I didn’t realize Jim Keller rejoined Intel and left again purportedly due to an internal argument about outsourcing.
|
# ¿ Jan 6, 2021 21:05 |
|
FuturePastNow posted:We don't know why he left or what he was working on. It is a mystery! https://www.reuters.com/article/us-intel-thirdpoint-exclusive-idUSKBN2931PS quote:Exclusive: Hedge fund Third Point urges Intel to explore deal options who knows but it wouldn't surprise me
|
# ¿ Jan 7, 2021 01:49 |
|
it's always fascinating to follow the implosion of a successful organization due to bean counters and office politickers taking over you have to give microsoft credit for being able to renew itself especially when you look at other big tech cos from its era
|
# ¿ Jan 7, 2021 02:24 |
|
obligatory
|
# ¿ Jan 7, 2021 04:03 |
|
there are 2-slot blower Ampere cards a lot of people use risers w/ cables to mount bigger cards anyway
|
# ¿ Jan 12, 2021 01:09 |
|
I have to wonder if there's any market for NUCs. My brother got a current gen one for casual gaming on the TV but gave up on it after realizing how anaemic its GPU capabilities were. He idly considered hooking up an eGPU to it but at that point, you're better off getting a normal desktop
|
# ¿ Jan 18, 2021 22:38 |
|
seems badquote:Users looking at our gaming results will undoubtedly be disappointed. The improvements Intel has made to its processor seem to do very little in our gaming tests, and in a lot of cases, we see performance regressions rather than improvements. If Intel is promoting +19% IPC, then why is gaming so adversely affected? The answer from our side of the fence is that Rocket Lake has some regressions in core-to-core performance and its memory latency profile.
|
# ¿ Mar 5, 2021 23:54 |
|
|
# ¿ May 9, 2024 10:36 |
|
lol I’m curious what you guys are doing are home that needs that much bandwidth
|
# ¿ Mar 8, 2021 05:33 |