|
The permission bureaucracy is the same when you are inside. But the big hurdle is usually the bad discoverability, just finding what you are looking for. You hope there is a wiki page that lists what roles you need. Otherwise, you have to know who to ask. Your job might be writing software for future architectures, but you have to know a guy who knows a guy who you can ask what permissions you need to access what unlisted windows share with the compiler nighlies.
|
#
?
Aug 3, 2023 10:18
|
|
|
|
| # ? May 26, 2024 15:58 |
|
|
mobby_6kl posted:Not at Intel but last time I got 1 (one) RSU. It was on top of the regular compensation but still it comes off as a joke. I think I would actually murder for $500k of RSUs though I'm assuming that it was a quarterly vestment over 4 years and due to rounding you wouldn't get your stock until the final quarter? 
|
|
#
?
Aug 3, 2023 13:41
|
|
|
Beef posted:The permission bureaucracy is the same when you are inside. But the big hurdle is usually the bad discoverability, just finding what you are looking for. You hope there is a wiki page that lists what roles you need. Otherwise, you have to know who to ask. Oh yeah finding stuff is really hard too. I usually find something similar then look at the references table for the 6 digit code for easier searching - although sometimes they are hidden, sometimes they show up but are locked, etc. But at least it makes it easier to pester our contact for something specific we know should exist! 
|
|
#
?
Aug 3, 2023 15:07
|
|
|
Beef posted:The permission bureaucracy is the same when you are inside. But the big hurdle is usually the bad discoverability, just finding what you are looking for. You hope there is a wiki page that lists what roles you need. Otherwise, you have to know who to ask. There was a system I needed access to that was controlled through a formal enterprise rights management thing with an approval chain. The application owner would get the access request, which included the employee's full details (department, title, role, etc) and a justification which was already approved by their first level manager. This application owner was so persnickety about the requests that she would deny them unless you followed the specific script she wrote. She denied my request because I didn't paste in the exact verbiage into the justification field, despite having met me face to face multiple times, knowing my role, and sitting at a cube 30 feet away. This was the factory organization, so nobody thought spending a week navigating that bureaucracy was surprising. Contrast that to an HR role, where the access process for getting EXTREMELY sensitive employee information (health records, home address, location data) that literally required written approval from Legal and an explicit business need would usually be fully completed within an hour or two. Another funny thing that happened was duplicate enterprise software licenses and deployments. Company has giant enterprise deployment of big dollar software (SAP, Oracle, Microsoft products, Salesforce). A department wants to begin using it or add features to it. Chronically underresourced IT requires them to jump through a bunch of hoops to make that happen. The department is impatient, so they end up just buying another license, hiring a third party integrator and standing up their own deployment. That's how you end up with the situation where people ask "why do we pay for 8 different production on-prem and cloud deployments of Sharepoint, some of which are literally running on a server underneath someone's desk?"
|
|
#
?
Aug 3, 2023 19:11
|
|
|
Permission control ran the gamut. In some cases I’ve seen sensitive documents and test plans shared on some random rear end windows share someone built from scraps and threw on the network and didn’t maintain
|
|
#
?
Aug 3, 2023 20:06
|
|
|
canyoneer posted:There was a system I needed access to that was controlled through a formal enterprise rights management thing with an approval chain. The application owner would get the access request, which included the employee's full details (department, title, role, etc) and a justification which was already approved by their first level manager. This application owner was so persnickety about the requests that she would deny them unless you followed the specific script she wrote. She denied my request because I didn't paste in the exact verbiage into the justification field, despite having met me face to face multiple times, knowing my role, and sitting at a cube 30 feet away. This was the factory organization, so nobody thought spending a week navigating that bureaucracy was surprising. Not a current or former intel employee, and not to say that the factory org isn't bureaucratic, but it can also be a problem of documented (and measured) routine process vs an ad-hoc or rarely-used one-off request. Someone is making sure that Karen in HR is processing her forms, because that's a major, textual part of her job description, and it happens every day. If you have someone who processes three requests a year... nobody's keeping an eye on that to make sure they're turning it around.
|
|
#
?
Aug 3, 2023 20:07
|
|
|
I’m still salty about some engineer chewing me out over email on my second day of work after I blasted a bunch of permission requests that were listed on a new hire wiki because saying “new hire on team X my responsibility is Y” isn’t sufficient with my manager CC’d even though that team literally existed only to build poo poo for us
|
|
#
?
Aug 3, 2023 20:20
|
|
|
Twerk from Home posted:It's not, which is why Intel's corporate strategy has always been to hire a thousand fresh college grads and lose most of them to Broadcom, Samsung, or other after a few years. They always insisted on the best of the best, top-shelf engineers that were heads above their peers... then paid in the absolute middle of the comp range. I left for a startup in SF. WhyteRyce posted:I’m still salty about some engineer chewing me out over email on my second day of work after I blasted a bunch of permission requests that were listed on a new hire wiki because saying “new hire on team X my responsibility is Y” isn’t sufficient with my manager CC’d even though that team literally existed only to build poo poo for us I took over this godawful perl script from someone in Texas, it would scrape the bug tracker and break down by cluster to provide a dashboard to mgmt, one-click to email all the folks who had bugs past a certain state. It was hacks upon hacks, just some of the worst code I've ever seen employed professionally, but it wasn't really mission-critical and it wasn't mine to fix so I just got it working well enough. Someone heard about it and I forwarded along the copy. I got back a blistering email about all the warnings that it spewed out to the console while updating, just "how dare you send me this code that emits any warning, it's not pristine" and... yeah I just had to let that one slide.
|
|
#
?
Aug 3, 2023 23:26
|
|
|
JawnV6 posted:They always insisted on the best of the best, top-shelf engineers that were heads above their peers... then paid in the absolute middle of the comp range. I left for a startup in SF. I once went on a recruiting trip with my boss to a UC. The hiring guidelines were 3.0 or higher. My boss was livid that we were getting all these 3.2 and 3.3 people and complaining loudly to the advisor we were working with about why the A students weren't lining up. I didn't know how to tell him top tier candidates weren't going to line up to do post-silicon chipset validation in a not exciting city for not FAANG pay That being said, if some borderline candidate managed to wrangle an interview, not have any major "absolutely do not hire" red flags, and the hiring manager starts hearing that the reqs are closing, you better believe they'll get hired WhyteRyce fucked around with this message at 01:32 on Aug 4, 2023 |
|
#
?
Aug 4, 2023 01:28
|
|
|
Hasturtium posted:This is all really interesting, thank you. Can you link to some resources so I can read more about this? search for 9780849337581 on the world's leading scientific paper delivery platform (use 2page mode imo) if you have specific questions fire away, you don't have to read the textbook but it explains how we got here from there vector processing has always been at the heart of the fastest chips, and keeping the beast fed (and keeping it from dealing with administrivia/t=1 thread loops/etc) has always been crucial. edit: section 7.3 is extremely extremely funni about how wrong all of that is on AI models Paul MaudDib fucked around with this message at 08:27 on Aug 4, 2023 |
|
#
?
Aug 4, 2023 07:38
|
|
|
.
Paul MaudDib fucked around with this message at 03:41 on Aug 14, 2023 |
|
#
?
Aug 4, 2023 07:47
|
|
|
Sometimes I wish there was a post silicon validation thread but mostly we’d probably just bitch about using tools and processes that were put in place before we were even out of college At my last job we were frozen using Tcl 8.4 because going to 8.5 would break things in ways we did not want to have to deal with. It was released in 2002.
|
|
#
?
Aug 4, 2023 08:04
|
|
|
priznat posted:Sometimes I wish there was a post silicon validation thread but mostly we’d probably just bitch about using tools and processes that were put in place before we were even out of college That’s just all big old companies, though especially regulated process oriented ones like manufacturing.
|
|
#
?
Aug 4, 2023 15:58
|
|
|
WhyteRyce posted:That being said, if some borderline candidate managed to wrangle an interview, not have any major "absolutely do not hire" red flags, and the hiring manager starts hearing that the reqs are closing, you better believe they'll get hired We had this grad student who was on the org chart for years because he kept deferring and the req would've evaporated if managers revoked the offer. JawnV6 fucked around with this message at 03:21 on Aug 11, 2023 |
|
#
?
Aug 6, 2023 17:45
|
|
|
Hasturtium posted:This is all really interesting, thank you. Can you link to some resources so I can read more about this? (latest bullshit)  recommended clips: general optimization / ue5 https://www.youtube.com/watch?v=NDEka3tBE1g&t=1943s royal core speculation https://www.youtube.com/watch?v=NDEka3tBE1g&t=5594s as usual, pay more attention to the guest than MLID, and remember this is just speculation and riffing. I really do think the idea of queue processing as a game engine is kinda prescient, and maybe that stuff scales out to some kind of micromailbox-server. scalable queue processing is a fairly proven problem, you can design some reasonably optimal mailbox topology for your world. edit: the end gets Weird Paul MaudDib fucked around with this message at 05:06 on Aug 8, 2023 |
|
#
?
Aug 8, 2023 04:36
|
|
|
https://downfall.page/ looks like a similar attack to that AMD info leak that came to light recently aw poo poo here we go again
|
|
#
?
Aug 8, 2023 20:56
|
|
|
repiv posted:https://downfall.page/ Not similar to zenbleed if that’s what you are comparing to 
|
|
#
?
Aug 9, 2023 03:39
|
|
|
repiv posted:https://downfall.page/ Our first mistake was teaching sand to think The second was asking it to peek forward in time
|
|
#
?
Aug 9, 2023 09:53
|
|
|
Potato Salad posted:Our first mistake was teaching sand to think The third mistake was to trust them with all of our secrets.
|
|
#
?
Aug 9, 2023 22:11
|
|
|
morris's three golden laws of computer security: do not own a computer, do not turn it on, and do not use it
|
|
#
?
Aug 10, 2023 02:43
|
|
|
Phoronix benchmarks the mitigation microcode in a handful of architectures: https://www.phoronix.com/review/intel-downfall-benchmarks
|
|
#
?
Aug 10, 2023 04:38
|
|
|
Dr. Video Games 0031 posted:Phoronix benchmarks the mitigation microcode in a handful of architectures: https://www.phoronix.com/review/intel-downfall-benchmarks do they have four arms and two heads? that was fast
|
|
#
?
Aug 10, 2023 22:14
|
|
|
repiv posted:https://downfall.page/ I saw "downfall.page" and thought it was going to be a "hitler reacts" site but with Intel.
|
|
#
?
Aug 10, 2023 22:50
|
|
|
Has anyone here used one of the IHS shims on a 12th or 13th gen CPU? Wondering whether they're actually needed in terms of the potential thermal failures.
|
|
#
?
Aug 10, 2023 23:49
|
|
|
SourKraut posted:Has anyone here used one of the IHS shims on a 12th or 13th gen CPU? Wondering whether they're actually needed in terms of the potential thermal failures. I thought about getting one for an alderlake but then thought eh, 
|
|
#
?
Aug 11, 2023 00:13
|
|
|
SourKraut posted:Has anyone here used one of the IHS shims on a 12th or 13th gen CPU? Wondering whether they're actually needed in terms of the potential thermal failures. I have the thermalright contact frame and it's easy to install. I have no idea if it's improving anything as I didn't want to spend the time to bench before and after.
|
|
#
?
Aug 11, 2023 01:34
|
|
|
I really wish Intel would start selling bare-die CPUs with precisely-machined shims. I'd pay a premium for it - I never chipped/killed any bare-die CPUs I ever owned.
|
|
#
?
Aug 11, 2023 01:39
|
|
|
priznat posted:I thought about getting one for an alderlake but then thought eh, Yeah, I haven't put one on my 12700K, but I'm putting a system together for a friend and remembered it, so was thinking to put it on now. phongn posted:I have the thermalright contact frame and it's easy to install. I have no idea if it's improving anything as I didn't want to spend the time to bench before and after. I'm guessing it's this one? https://www.amazon.com/Thermalright-Contact-Retrofit-17XX-BCF-Generation/dp/B0B811J7D9?th=1
|
|
#
?
Aug 11, 2023 06:52
|
|
|
I've seen plenty of people fix bad thermals by installing that, but it's one of those things that's very dependent on the motherboard and cooler you use. Some boards cause the IHS to flex more than others, and some coolers respond more poorly to the flexing than others (and some may conceivably benefit, it depends). If your thermals are fine as it is, then I wouldn't bother.
|
|
#
?
Aug 11, 2023 07:24
|
|
|
It’s beautiful(ly tacky!) https://videocardz.com/newz/colorfire-b760m-meow-motherboard-gives-paws-for-thought Just need to equip the Yeston cat GPU with it.
|
|
#
?
Aug 11, 2023 16:47
|
|
|
I have something of a history question: what were server CPUs like in the era before multi-core CPUs, in terms of how different they were from desktop CPUs?
|
|
#
?
Aug 12, 2023 02:10
|
|
|
gradenko_2000 posted:I have something of a history question: what were server CPUs like in the era before multi-core CPUs, in terms of how different they were from desktop CPUs? Are you talking about x86 or non-x86 machines? Also, a plenty of server-class machines had multi-socket motherboards, so you had multiple cores available anyway across the 2 or 4 sockets. The Pentium Pro was the first really different x86 server chip: https://en.wikipedia.org/wiki/Pentium_Pro. It was really expensive and just a bigger chip in every way possible, and paved the way for Pentium 2 and PIII. Just significantly better IPC, bigger caches, bigger feature sizes. If you want somebody to talk about VAXen or DEC Alphas or POWER or something, I hope someone else chimes in. Twerk from Home fucked around with this message at 02:31 on Aug 12, 2023 |
|
#
?
Aug 12, 2023 02:26
|
|
|
Got me thinking of the Pentium-II slot1 cartridges now And the massive pentium II Pro ones, they were like the P2 with a fivehead.
|
|
#
?
Aug 12, 2023 02:33
|
|
|
gradenko_2000 posted:I have something of a history question: what were server CPUs like in the era before multi-core CPUs, in terms of how different they were from desktop CPUs? There's a really lovely thread right on these forums, including a stretch on the first page where I embarrass myself trying to talk about Power(PC). There's some interesting info in there, and asking this question there would likely net you some solid, interesting answers.
|
|
#
?
Aug 12, 2023 02:52
|
|
|
gradenko_2000 posted:I have something of a history question: what were server CPUs like in the era before multi-core CPUs, in terms of how different they were from desktop CPUs? On the x86 side? Early on, not very. https://en.m.wikipedia.org/wiki/Xeon Lots of glue on the mobos and chip/card. The high end RAS stuff lived on the mobo. Once the memory controller moved on board it started to diverge more.
|
|
#
?
Aug 12, 2023 04:38
|
|
|
gradenko_2000 posted:I have something of a history question: what were server CPUs like in the era before multi-core CPUs, in terms of how different they were from desktop CPUs?  Then there's the more normal Supermicro X13QEH+ which, despite its form-factor, is fairly standard and has one CPU connected to the PCH (which is the equivalent to the old South Bridge, plus some added circuitry for the Management Engine as well as System Attestation):  Finally, there's the older style of boards where the CPUs are connected to the North Bridge (which contains the memory controller and links to the daughterboard bus), like the Supermicro P3TDL3:  There are also older much more proprietary designs, like the Compaq SystemPro (which probably got its start in life when Compaq bought DEC, who'd done multi-processing motherboards since 1982?) - but it's much harder to find information about those. BlankSystemDaemon fucked around with this message at 11:27 on Aug 12, 2023 |
|
#
?
Aug 12, 2023 11:09
|
|
|
gradenko_2000 posted:I have something of a history question: what were server CPUs like in the era before multi-core CPUs, in terms of how different they were from desktop CPUs? Before Intel ate everyone's lunch (cfr. Rise of the Killer Micro) on the server side, there were bespoke processor architectures for supercomputers, mainframes, minicomputers and workstations. Modern CPUs look more like evolutions of those old server processors than what was the desktop equivalent at the time, but what used to be a different board or rack has migrated on-die. Microcomputers were laughably under-powered at the time. You see this extinction event in CS literature, with a rich biotope of weird and interesting species (Transputer, Lisp machines, Connection Machine, dataflow processors, ...) before and boringly uniform ecosystem of microprocessor CPU after.
|
|
#
?
Aug 12, 2023 17:51
|
|
|
Beef posted:Before Intel ate everyone's lunch (cfr. Rise of the Killer Micro) on the server side, there were bespoke processor architectures for supercomputers, mainframes, minicomputers and workstations. Modern CPUs look more like evolutions of those old server processors than what was the desktop equivalent at the time, but what used to be a different board or rack has migrated on-die. Microcomputers were laughably under-powered at the time. The last word in history hasn't been written yet.
|
|
#
?
Aug 12, 2023 18:37
|
|
|
A chinese reviewer has allegedly gotten their hands on an early 14700K sample and benchmarked it: https://videocardz.com/newz/intel-core-i7-14700k-tested-and-compared-to-13700k-up-to-20-7-faster-multi-core-but-needs-30-more-watts Seems very much like you'd expect from a refresh.
|
|
#
?
Aug 12, 2023 23:57
|
|
|
|
| # ? May 26, 2024 15:58 |
|
|
BlankSystemDaemon posted:Mainframes still run a different ISA, and ARM and RISC-V are steadily rising. Processor-architecture, not ISA. (e.g. Intel could run a Cray XMT- or SPARC-style 16-way SMT barrel-processor with x86. ) The rise of the killer micro has nothing to do with intrinsic qualities of any ISA and everything to do with fab improvements and economies of scale.
|
|
#
?
Aug 13, 2023 00:44
|
|