|
Eletriarnation posted:I like the e-cores, but they need to go in the other direction with it too. Give me a successor to the Compute Stick with an N100 inside. I'd want to see that too but the sticks look pretty dead. However some mini-pcs with the N305 are popping up. The Beelink EQ12 for example, though these tend to be on the expensive side compared to more generic versions: https://www.aliexpress.com/item/1005005374184908.html There's even a review I found, no English subs but it's mostly self-explanatory. Only the CB23 is bizarrely low, only 4.3 MP ratio for some reason, but otherwise looks pretty cool. https://www.youtube.com/watch?v=6jtZvTI1TJ8 E: So it actually does better than the 1+4 Pentium 8505 in CB, slightly higher score at just over half the wall power (62W in another video). Still, pretty bizarre scaling, and nobody looks at actual core power and frequencies. I guess I'll have to get one myself and see if Chips and Cheese doesn't. mobby_6kl fucked around with this message at 14:37 on Apr 3, 2023 |
#
?
Apr 2, 2023 09:46
|
|
|
|
| # ? May 23, 2024 15:49 |
|
|
Eletriarnation posted:I like the e-cores, but they need to go in the other direction with it too. Give me a successor to the Compute Stick with an N100 inside. It's extremely amusing to me that the only thing anyone here is asking for from Intel are things they can plug in via USB to make their existing computers better.
|
|
#
?
Apr 4, 2023 06:14
|
|
|
av1 encoder/decoder on a stick could still make them some cheap profit.
|
|
#
?
Apr 4, 2023 06:51
|
|
|
and also take the wind out of the sails of nvidia, so win-win
|
|
#
?
Apr 4, 2023 09:27
|
|
|
BlankSystemDaemon posted:EDIT: Hilariously, Intel still treats Gordon Moore like some kind of god, and keep praying to his shrine despite the fact that everyone recognizes that both Dennard scaling and Moore's law (the part about economics) are dead as the dodo. "im not owned! im not owned!!", i continue to insist, as my transistors slowly shrink and transform into space heaters
|
|
#
?
Apr 5, 2023 19:22
|
|
|
meteor lake will apparently have L4 cache cache rules everything around me
|
|
#
?
Apr 12, 2023 17:38
|
|
|
Klyith posted:meteor lake will apparently have L4 cache EDIT: Ian did a five-year retrospective on it, and even from a performance point it's hard to judge whether it makes sense. BlankSystemDaemon fucked around with this message at 18:15 on Apr 12, 2023 |
|
#
?
Apr 12, 2023 18:09
|
|
|
BlankSystemDaemon posted:Intel tries this exact stunt every so often, and it never works out. I mean, it does work well for some workloads. It isn't always cost effective though. With Meteor Lake being tile based (and future parts being 3D stacked), perhaps the cost issue has changed? The Xeon E3-1285L V4 in my HTPC can push 480 points in the little CPU-Z single core test, which is like... lower end Zen 2/Coffee Lake/Comet Lake territory, lol.
|
|
#
?
Apr 12, 2023 18:22
|
|
|
New Zealand can eat me posted:It's extremely amusing to me that the only thing anyone here is asking for from Intel are things they can plug in via USB to make their existing computers better. Not exactly... I want something to plug in via HDMI to make my TV better. Right now that's an i3-5010 NUC stuck to the back with a lot of Velcro, so it could use an update.
|
|
#
?
Apr 12, 2023 18:29
|
|
|
BlankSystemDaemon posted:Intel tries this exact stunt every so often, and it never works out - by which I mean that it leaves users without an upgrade path, because there's too long between each iteration of Intel having a go at it, and by the time they try again, you're basically build an entire new system anyway. I guess you could say Intel's mouth is writing checks its body can't cache 
|
|
#
?
Apr 12, 2023 18:42
|
|
|
I don't know why Intel just doesn't figure out a way to mate their high-end consumer CPUs with HBM for a HEDT-esque upgrade. Obviously not as much as is on Sapphire Rapids, but "enough."
|
|
#
?
Apr 12, 2023 19:02
|
|
|
Intel exiting the server box business: https://www.servethehome.com/breaking-intel-exiting-the-server-business-selling-to-mitac/ Likely makes business sense as a lot of things have standardized lately and lead to smaller margins. Intel def trimming a lot of smaller business units lately.
|
|
#
?
Apr 12, 2023 19:18
|
|
|
BlankSystemDaemon posted:Intel tries this exact stunt every so often, and it never works out - by which I mean that it leaves users without an upgrade path, because there's too long between each iteration of Intel having a go at it, and by the time they try again, you're basically build an entire new system anyway. That is a very weird take on this technology, imo. On-package L4 eDRAM was mostly an Apple request. Apple wanted an iGPU fast enough to replace dGPUs in some notebooks and also not suck terribly in dual-graphics machines. To pull that off, Intel needed to offer a lot more memory performance - they'd been designing and sizing their iGPUs around the bandwidth available as a side effect of having a CPU. Intel offered the L4 cache concept to Apple, they accepted, and many generations of Intel laptop chips shipped with the L4 and a much bigger iGPU. So, for the main user of this tech, there actually was a regular upgrade cadence. It was a success, it stuck around a long time and made two companies a lot of money. The wider PC market didn't care as much, but it was the closest Apple could get to what it really wanted, and they lived with it for many years before going off to do their own thing. Intel seems to have stopped offering these products since Apple left, which is not surprising as Apple really was the main driver of demand, but I guess they've figured out a new application. I don't think users will have to worry about having no upgrade path, at the end of the day it's just a performance feature and if Intel wants people to buy new product generations they have to figure out how to offer better performance than the last gen. Whether that involves a L4 eDRAM cache or not is immaterial to the user.
|
|
#
?
Apr 12, 2023 21:01
|
|
|
gradenko_2000 posted:I guess you could say Intel's mouth is writing checks its body can't cache
|
|
#
?
Apr 13, 2023 03:21
|
|
|
BIG HEADLINE posted:I don't know why Intel just doesn't figure out a way to mate their high-end consumer CPUs with HBM for a HEDT-esque upgrade. Obviously not as much as is on Sapphire Rapids, but "enough." It's effectively impossible if you're trying to run their poo poo at "high-end consumer" clockspeeds while also keeping all of the HBM on top of it from becoming molten slag.
|
|
#
?
Apr 14, 2023 02:11
|
|
|
Wonder what EVGA has been up to? Making an $800 Z790 motherboard https://www.evga.com/products/product.aspx?pn=121-RL-E799-KR
|
|
#
?
Apr 21, 2023 09:50
|
|
|
Alright, we're here. VRM fans.
|
|
#
?
Apr 21, 2023 11:35
|
|
|
That's not exactly new.
|
|
#
?
Apr 21, 2023 12:12
|
|
|
Wow does it actually have finned heatsinks? They are ultra rare these days.
|
|
#
?
Apr 21, 2023 12:15
|
|
|
Beef posted:Alright, we're here. VRM fans. As in, actual metal heatsinks and not some dumb looking piece of plastic?
|
|
#
?
Apr 21, 2023 17:56
|
|
|
did the old 350+ watts Pentium systems not have vrm heat spreaders?
|
|
#
?
Apr 21, 2023 23:50
|
|
|
Potato Salad posted:did the old 350+ watts Pentium systems not have vrm heat spreaders? They didn’t need them, as the electrolytic capacitors were designed to explode long before the overtaxed VRM’s gave up.
|
|
#
?
Apr 22, 2023 01:36
|
|
|
intel didn't really had high TDP consumer CPUs until the pentium 4
|
|
#
?
Apr 22, 2023 01:49
|
|
|
Palladium posted:intel didn't really had high TDP consumer CPUs until the pentium 4 Laughing at those garbage 72W t-birds from a 30W coppermine. 72W for a ridiculous overclocking CPU seems quaint but think of that like 72 W for a single core
|
|
#
?
Apr 22, 2023 02:04
|
|
|
Potato Salad posted:did the old 350+ watts Pentium systems not have vrm heat spreaders? the highest desktop draw parts in the Pentium flagship era were 130w, which was on 775. boards with VRM heatsinks did exist, especially later in the 775 run, but they were mostly a gimmick iirc. many didnt even have thermal pads, lol the highest power socket 478 parts were 103w, and ran just fine on boards that looked like this.  i do remember there being some 478 high end overclocker boards with heatsinks i think?
|
|
#
?
Apr 22, 2023 03:23
|
|
|
Oh hey the first desktop I ever owned had a board that looked a lot like that, the ASUS P4C800-E or something like that. Ran a Pentium 4 at 3.6GHz and a Pentium M at 2.56 (with a socket adapter) for a few years each. It also ran Skyrim and WoW: Legion surprisingly well for a single core with an AGP graphics card back in 2016, so yeah I can't ding it's longevity.
|
|
#
?
Apr 22, 2023 03:40
|
|
|
yeah its funny because nobody gave a hoot about VRM quality or temperatures during the era when CPUs were actually worth overclocking from a value PoV
|
|
#
?
Apr 22, 2023 03:42
|
|
|
Cygni posted:the highest desktop draw parts in the Pentium flagship era were 130w, which was on 775. boards with VRM heatsinks did exist, especially later in the 775 run, but they were mostly a gimmick iirc. many didnt even have thermal pads, lol How much power did those draw when people got them to 8ghz on liquid nitrogen though?
|
|
#
?
Apr 22, 2023 03:48
|
|
|
The AMD Quad FX motherboard had these little VRM fans back in 2006: Some other Asus mobos from that era used them too.
|
|
#
?
Apr 22, 2023 05:51
|
|
|
Beef posted:Alright, we're here again. VRM fans. ftfm
|
|
#
?
Apr 22, 2023 08:58
|
|
|
FuturePastNow posted:The AMD Quad FX motherboard had these little VRM fans back in 2006: that probably has an actual sales figure counted on 1 hand
|
|
#
?
Apr 22, 2023 09:21
|
|
|
Palladium posted:that probably has an actual sales figure counted on 1 hand Slower than a QX6700 for twice the power and twice the price? Very exciting
|
|
#
?
Apr 22, 2023 13:23
|
|
|
FuturePastNow posted:Slower than a QX6700 for twice the power and twice the price? Very exciting at least it's not glued together like those Intels heh 
|
|
#
?
Apr 22, 2023 17:43
|
|
|
FuturePastNow posted:Slower than a QX6700 for twice the power and twice the price? Very exciting i don't think we will see a $180 new cheapest CPU (E6300) wiping the floor of a $1000 old flagship (FX-60) before a monstrous 75% OC ever again
|
|
#
?
Apr 23, 2023 10:06
|
|
|
Going to upgrade my home server/NAS CPU to improve Plex transcoding performance. I specifically am going to upgrade to an Intel chip to take advantage of the Plex hardware transcoding + quick sync video offered by Intel. This server won't have a GPU, so whatever processor I end up with needs an iGPU. Transcoding video is the only demand that is realistically placed on the system. I'm doing some research and it feels like there's a ton of options. I was expecting to land on some kind of 12th gen i5 or even i3, but I'm curious what the collective mind thinks is a good option. What would you put in? I'm running win10, does 12th gen even make sense to use without win11 for the e-cores? Is the UHD 770 iGPU significantly better than the 730 on the lower end chips? Budget is around $300CAD ($220 USD). Thanks goons. VelociBacon fucked around with this message at 17:34 on Apr 23, 2023 |
|
#
?
Apr 23, 2023 17:31
|
|
|
Palladium posted:i don't think we will see a $180 new cheapest CPU (E6300) wiping the floor of a $1000 old flagship (FX-60) before a monstrous 75% OC ever again To be fair to the FX-60, the E6300 wiped the floor with Intel's previous flagship, too
|
|
#
?
Apr 23, 2023 18:43
|
|
|
VelociBacon posted:Is the UHD 770 iGPU significantly better than the 730 on the lower end chips? If all you care about on the GPU is the video de/encode, there's no difference.
|
|
#
?
Apr 23, 2023 19:03
|
|
|
Klyith posted:If all you care about on the GPU is the video de/encode, there's no difference. Thanks. I actually ended up finding an i5-11600k for $150 off, ended up being $199 so I just jumped on that and I think it'll be plenty.
|
|
#
?
Apr 23, 2023 19:18
|
|
|
https://twitter.com/VideoCardz/status/1650077328146325504 Meteor Lake lookin pretty wild. The L4 cache will be 3d stacked beneath the 4 dies and usable by any of them it seems. Supposedly the iGPU uplift will be pretty substantial from it. Also... what in the world is with IO die having its own 2 crestmont cores? 
|
|
#
?
Apr 23, 2023 20:18
|
|
|
|
| # ? May 23, 2024 15:49 |
|
|
Cygni posted:Supposedly the iGPU uplift will be pretty substantial from it. It was. Broadwell-C is calling from 2014
|
|
#
?
Apr 23, 2023 20:27
|
|
