Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us $3,400 per month for bandwidth bills alone, and since we don't believe in shoving popup ads to our registered users, we try to make the money back through forum registrations.
«561 »
  • Post
  • Reply
Space Gopher
Jul 31, 2006
BLITHERING IDIOT

Palladium posted:

If we going in the all +12V direction, why not just have PCI-E 8-pin replace the bulky 24-pin on the mobo side while standardizing all modular cables on the PSU end FFS.

You need a few extra pins if you want "ATX but 12V only" behavior - one for standby power, one for the power-on signal from the motherboard, and one for the "power good" signal from the PSU that tells the host system it's safe to start up/it needs to reset because of a power fault. It's a lot less than a standard ATX setup, but you can't get away with a peripheral connector that depends on the motherboard for handling all those functions.

Adbot
ADBOT LOVES YOU

bull3964
Nov 18, 2000

DO YOU HEAR THAT? THAT'S THE SOUND OF ME PATTING MYSELF ON THE BACK.




Malcolm XML posted:

why not full 48v like servers. highest efficiency, smallest cables

Pfft, I'll do you one better. Pretty much everything in our homes now runs on DC aside from major appliances. Let's just convert home wiring over to DC and do the conversion at the house level!

Think of how much smaller everything could be without DC power supplies!

Paul MaudDib
May 2, 2006

"Tell me of your home world, Usul"


Malcolm XML posted:

why not full 48v like servers. highest efficiency, smallest cables

it seems like being able to feed high-current devices like GPUs with 48V (or, pick 40v since it's an integer factor of 120V?) would be pretty nifty. From what I remember about power supply circuitry, that should increase the efficiency, right? Unless the present 12V is some kind of optimum for the VRM input mosfets or something?

(high current devices like comet lake, go for that high-voltage FIVR )

The power they're pushing the 10900K to is just tremendous, the 9900K can push 200W at full load, this goes from 4.7 all-core to 4.8 and adds two more cores. Board partners specced for like 300-350 watts supposedly lmao. If people are trying to hit above-5GHz all core they're going to need it. just ship a factory direct-die-cooling model already you cowards!

I kinda have a hunch that 10C is going to become the breaking point of thermals and the resulting throttling is going to start reducing clocks. We'll see how the package changes ended up affecting thermals, that's the money question. Like, if you are thinking seriously about this part you might seriously want to budget for a professional delidding (about $60+sh) and potentially even running one of those direct die support frames. Thermals are gonna be pretty ugly.

Paul MaudDib fucked around with this message at 03:46 on May 6, 2020

Malcolm XML
Aug 8, 2009

I always knew it would end like this.


Paul MaudDib posted:

it seems like being able to feed high-current devices like GPUs with 48V (or, pick 40v since it's an integer factor of 120V?) would be pretty nifty. From what I remember about power supply circuitry, that should increase the efficiency, right? Unless the present 12V is some kind of optimum for the VRM input mosfets or something?

(high current devices like comet lake, go for that high-voltage FIVR )

(the power they're pushing the 10900K to is just tremendous, the 9900K can push 200W at full load, this goes from 4.7 all-core to 4.8 and adds two more cores. Board partners specced for like 300-350 watts supposedly lmao. If people are trying to hit above-5GHz all core they're going to need it. just ship a factory direct-die model already you cowards!)

vicor has modules they're touting as being able to do this, mostly for """ai""" processors (DSPs), they claim a huge reduction in output resistance losses: http://www.vicorpower.com/documents...iency-VICOR.pdf

Malcolm XML fucked around with this message at 03:45 on May 6, 2020

Malcolm XML
Aug 8, 2009

I always knew it would end like this.


most of a motherboard/gpu is dedicated to power distribution so u could have really tiny gpus with more efficient power handling

Cygni
Nov 12, 2005

raring to post



Paul MaudDib posted:

(the power they're pushing the 10900K to is just tremendous, the 9900K can push 200W at full load, this goes from 4.7 all-core to 4.8 and adds two more cores. Board partners specced for like 300-350 watts supposedly lmao. If people are trying to hit above-5GHz all core they're going to need it. just ship a factory direct-die model already you cowards!)

The anandtech review of the 9900k had this handy chart. This was at 4.7ghz from the register values (not using a current clamp):



So 186w is about what I would expect to be dealing with at 4.7 all core (10900k's all core is 4.8, so not fully 1 to 1). Going for 5ghz all core all the time will be... more than that. The PL2 for the 9900k for reference is 210w.

B-Mac
Apr 21, 2003
I'll never catch "the gay"!

Prime95 Small FFT will pull roughly 210W package power with my stock clock undervolted 9900K, but that’s obviously an unreasonable load. Blender and ASUS realbench are closer to 150W. The 10900K overclocked is gonna have hilarious power draw.

Shaocaholica
Oct 29, 2002

Fig. 5E


Space Gopher posted:

You need a few extra pins if you want "ATX but 12V only" behavior - one for standby power, one for the power-on signal from the motherboard, and one for the "power good" signal from the PSU that tells the host system it's safe to start up/it needs to reset because of a power fault. It's a lot less than a standard ATX setup, but you can't get away with a peripheral connector that depends on the motherboard for handling all those functions.

How are consoles setup? The xbox one x literally only has 12Vx1 and GNDx1 from its PSU.

sincx
Jul 13, 2012

What actually transpires beneath the veil of an event horizon? Decent people shouldn't think too much about that.

bull3964 posted:

Pfft, I'll do you one better. Pretty much everything in our homes now runs on DC aside from major appliances. Let's just convert home wiring over to DC and do the conversion at the house level!

Think of how much smaller everything could be without DC power supplies!

Edison was right after all

Paul MaudDib
May 2, 2006

"Tell me of your home world, Usul"


MSI discusses Intel Comet Lake-S processors binning and overclockability

203W stock, 347W at 5.1 all-core.

I think I remember reading that the 10700 series will be 10C with disabled cores. It also looks like the 10900K/KF are getting the cream of the crop as far as binning, an "average" 10900K should do 5.1 all-core at about 1.25V.

Cygni
Nov 12, 2005

raring to post





Pretty much as expected vs the 9900K. Considering the 3900X pulls like 215w from the wall, the 10900K with (probably?) similar performance pulling 276w at the wall I guess really isnt a disaster. Certainly not ideal though. And obvi wait for reviews.

The preorders for motherboard are up. Yes, despite the fact that the review embargo is weeks away. And yes, despite the CPU pre orders not being up yet. Someone please regulate these companies.

https://www.newegg.com/p/pl?N=10000...490&Order=PRICE

Cygni fucked around with this message at 18:52 on May 6, 2020

taqueso
Mar 8, 2004









Fun Shoe

Paul MaudDib posted:

it seems like being able to feed high-current devices like GPUs with 48V (or, pick 40v since it's an integer factor of 120V?) would be pretty nifty. From what I remember about power supply circuitry, that should increase the efficiency, right? Unless the present 12V is some kind of optimum for the VRM input mosfets or something?

The big plus with 48V is that you get more power at the same current. Current drives wire size and component size, so you can have smaller cables or use more power with the same cables. A good DC-DC converter is going to be 90%+ efficient, and parts are available for 48V with the same or at least very similar performance characteristics. I don't think there would be a downside beyond breaking compatibility with everything.

48V is a good spot since it is not considered high voltage by regulatory bodies, so it is common and there are a lot of off the shelf parts.

Cygni
Nov 12, 2005

raring to post



Looking through these Z490 motherboards, I'm not sure if it was better or worse when consumers didn't hyper focus on VRMs. All of the big 4 have figured out that they can put incredibly overboard VRMs on everything and people will buy them for huge prices, when they absolutely don't need them. So many of these boards destined for 6 core CPUs are just comically overbuilt.

For example, lets check out A Board:

Full ATX, 4 DIMM
True 10+2 VRM, ISL69269 controller (not sure on the fets/stages, but lets just assume SiC651A 50A DrMOS cause thats prolly what it is)
VRM heatsinks (not huge ones, but look to be functional and not for show)
2.5G Intel networking
2x M.2 + 1x M.2 e-key for wifi
ALC1220 audio
6 SATA


Like... thats the audio you want, the NIC you want, a good M.2 config, a very good VRM. Gotta be premium, right? On Z390, thats probably a $250+ board. Hell, there weren't many true 10 on more phase boards at all on Z390. Well, thats a $159 board on Z490.

Obviously wait for reviews and such, but I guess what I'm saying is I am real skeptical that you really need any of these $300+ boards even if you getting the 10 core.

eames
May 9, 2009



Paul MaudDib posted:

MSI discusses Intel Comet Lake-S processors binning and overclockability

203W stock, 347W at 5.1 all-core.

I think I remember reading that the 10700 series will be 10C with disabled cores. It also looks like the 10900K/KF are getting the cream of the crop as far as binning, an "average" 10900K should do 5.1 all-core at about 1.25V.

Seems like WCCF made a mistake there or the slide is wrong, because it clearly states 203W stock/240W at 51x. The 347W number is total system power.
I’m sure just was a honest mistake, WCCF is not known for manufacturing sensational content... right?

Those numbers are high but better than I expected. I suspect they would still go up into the 300s with synthetic workloads, because that’s just Cinebench we are seeing here.

It was Computerbase that speculated that all -K SKUs use 10C die. That die seems to be the only one that gets soldered TIM, the lower Z-height and the thicker IHS to make up for it. It could make the six and eight core enthusiast models quite decent overclocking candidates due to the lower heat density.

Cygni posted:

Obviously wait for reviews and such, but I guess what I'm saying is I am real skeptical that you really need any of these $300+ boards even if you getting the 10 core.

Yeah, I totally agree. The manufacturers made mistakes with some Z370 boards using „cosmetic heatsinks“, so VRMs and cooling became a big topic, even though most boards still did ok.
Z390 already fixed that for the most part and now Z490 seems to double down on it because by now the topic has become a huge (if not the deciding) factor in reviews and thus marketing. One can only hope that overall MB quality doesn’t suffer from this development.

eames fucked around with this message at 06:26 on May 7, 2020

Palladium
May 8, 2012


Cygni posted:

Looking through these Z490 motherboards, I'm not sure if it was better or worse when consumers didn't hyper focus on VRMs. All of the big 4 have figured out that they can put incredibly overboard VRMs on everything and people will buy them for huge prices, when they absolutely don't need them. So many of these boards destined for 6 core CPUs are just comically overbuilt.

For example, lets check out A Board:

Full ATX, 4 DIMM
True 10+2 VRM, ISL69269 controller (not sure on the fets/stages, but lets just assume SiC651A 50A DrMOS cause thats prolly what it is)
VRM heatsinks (not huge ones, but look to be functional and not for show)
2.5G Intel networking
2x M.2 + 1x M.2 e-key for wifi
ALC1220 audio
6 SATA


Like... thats the audio you want, the NIC you want, a good M.2 config, a very good VRM. Gotta be premium, right? On Z390, thats probably a $250+ board. Hell, there weren't many true 10 on more phase boards at all on Z390. Well, thats a $159 board on Z490.

Obviously wait for reviews and such, but I guess what I'm saying is I am real skeptical that you really need any of these $300+ boards even if you getting the 10 core.

The only bad thing is mobo makers getting their profit margins squeezed too hard by people who aren't buying boards catering to e-peen waving Youtubers.

Twerk from Home
Jan 17, 2009

This avatar brought to you by the 'save our dead gay forums' foundation.


I've noticed a ton of new 2020 PC models are getting introduced with 8th gen Core CPUs. What's going on here? Intel list pricing isn't any lower for 8th vs 9th or 10th, and there are both 10nm and 14nm 10th gen parts available so it seems like this couldn't be purely a supply issue. With identical pricing, why not take the extra couple hundred megahertz from using new parts rather than 2 year old ones?

For example, the just updated Surface Go 2 got an 8th gen Core m3, the newly refreshed 13" Macbook Pro got an 8th gen i5, and a ton of commodity laptops seem to be sticking to 8th gen parts for the long run.

Inept
Jul 8, 2003

3 olives has a payment plan with his phone company for his cell phone

Likely Intel is cutting OEMs some deals. No one pays list price.

Anime Schoolgirl
Nov 28, 2002






The 4+1 core Lakefield is the only new Atom-based core design so far, right? That lineup has been rather quiet lately, with the most recent releases being slightly upclocked processors showing up in 2019.

BobHoward
Feb 13, 2012

The only thing white people deserve is a bullet to their empty skull


Probably a lot of it is that they aren’t going to move all their high volume fabs to 10nm, ever. For Intel 10 has become a stopgap for a narrow range of products until they can launch their 7nm node.

Cygni
Nov 12, 2005

raring to post



Anime Schoolgirl posted:

The 4+1 core Lakefield is the only new Atom-based core design so far, right? That lineup has been rather quiet lately, with the most recent releases being slightly upclocked processors showing up in 2019.

Yeah, the Tremont Atom stuff was supposed to be on 10nm so... yeah. Never really happened. There were some parts made for 5G towers and stuff called Snow Ridge, but they werent for public consumption.

Beyond Lakefield, the rumor is that Intel's Alder Lake CPUs are going to have a combo of 8 Golden Cove big cores and 8 Gracemont Atom cores, but we shall see.

icantfindaname
Jul 1, 2008




BobHoward posted:

Probably a lot of it is that they aren’t going to move all their high volume fabs to 10nm, ever. For Intel 10 has become a stopgap for a narrow range of products until they can launch their 7nm node.

Why do people think 7nn will be easier than 10? Not a diss, legitimately curious how they could struggle so much with 10 and then be able to do 7 fine and on schedule

SwissArmyDruid
Feb 14, 2014



Allegedly, Intel's woes came down to their inability to find a suitable pellicle for EUV, and now they have one that will be suitable for 5/7/10.

That's what they say, at least.

Khorne
May 1, 2002

Goonstone Champ x2

icantfindaname posted:

Why do people think 7nn will be easier than 10? Not a diss, legitimately curious how they could struggle so much with 10 and then be able to do 7 fine and on schedule
Everyone is transitioning to EUV now. 10 was initially an ambitious not-EUV process. They opted to move to 7 on EUV instead of investing a ton more money into getting 10 on EUV up to par.

It's ultimately a good call, because their 14nm process remained competitive in most segments.

Malcolm XML
Aug 8, 2009

I always knew it would end like this.


SwissArmyDruid posted:

Allegedly, Intel's woes came down to their inability to find a suitable pellicle for EUV, and now they have one that will be suitable for 5/7/10.

That's what they say, at least.

Intel doesn't use euv for 10 though

WhyteRyce
Dec 30, 2001



There are all sorts of plausible sounding theories from all sorts of places so the real answer is probably just hell dimension

DrDork
Dec 29, 2003
commanding officer of the Army of Dorkness

Malcolm XML posted:

Intel doesn't use euv for 10 though

Yeah, I think that was the entire problem.

In any event, they seem confident about 7nm. We'll see if they can deliver, though.

Cygni
Nov 12, 2005

raring to post



Heres the mask breakdown for Comet Lake, looks like only 2 distinct masks this time (a soldered 10 core and a 6 core with TIM) vs 4 with Coffee Lake R.

https://twitter.com/momomo_us/statu...705650869137409

Anime Schoolgirl
Nov 28, 2002






WhyteRyce posted:

There are all sorts of plausible sounding theories from all sorts of places so the real answer is probably just hell dimension
from what i've been reading on semiengineering and the like the problem was very likely that intel banked on the wrong materials to use at the metal layers and tsmc used the correct ones

Cygni posted:

Heres the mask breakdown for Comet Lake, looks like only 2 distinct masks this time (a soldered 10 core and a 6 core with TIM) vs 4 with Coffee Lake R.

https://twitter.com/momomo_us/statu...705650869137409
Intel finally doing core harvesting on non-4 core SKUs, nice

Anime Schoolgirl fucked around with this message at 00:32 on May 9, 2020

eames
May 9, 2009



The buildzoid video on selective hyperthreading is quite interesting.

The idea, as he explains it, is that it allows you to pick up extra overclocking headroom by selectively disabling hyperthreading on the "weak" cores of a CPU. 10C/16T at 5.2 GHz will usually be faster than 10C/20T at 5.1 GHz, so disabling HT on the right cores can be a overall performance gain.

Paul MaudDib
May 2, 2006

"Tell me of your home world, Usul"


Can’t you set a per core multiplier anyway, so you could just run that one core at 50x and the rest at 51x (lets say)? Why wouldn’t that be better than turning off HT on the core?

eames
May 9, 2009



Paul MaudDib posted:

Can’t you set a per core multiplier anyway, so you could just run that one core at 50x and the rest at 51x (lets say)? Why wouldn’t that be better than turning off HT on the core?

I'm not sure if Z490 can do that. I believe the newer Intel HEDT platforms can do it and I know Coffee Lake on Z370 can't.

Selective HT might still be better because the effect of Intel's SMT implementation is negligible (sometimes even negative) for some workloads while higher frequency always improves performance. This seems to be especially true past a certain thread count. To illustrate the point, you'd probably rather run the average 2020 game engine on a 5 GHz 32C/32T CPU than a 4.8 GHz 32C/64T CPU.
Buildzoid's whole explanation is aimed at video game titles and benchmarks that stop scaling with threads at some point. I imagine Ahmdal's law also ties into that.

Your approach — if technically possible on Z490 — makes sense for workloads that show perfectly linear scaling. Those workloads tend to run best on HEDT platforms which already have that feature, so maybe the engineers had that in mind.

BobHoward
Feb 13, 2012

The only thing white people deserve is a bullet to their empty skull


icantfindaname posted:

Why do people think 7nn will be easier than 10? Not a diss, legitimately curious how they could struggle so much with 10 and then be able to do 7 fine and on schedule

The leading theory on the 10nm problems is that Intel management got high on their own supply, greenlit too many risky new ideas at the 10nm node, and didn't act decisively to de-risk it when problems cropped up, since careers were staked on it succeeding.

I don't know why 7nm would be better, but if they think 7nm is on track to be ready soon, it makes sense for them to move on from ramping up 10nm capacity. It costs incredible amounts of money to build new or convert old fabs for a new node. Intel has lots of manufacturing capacity to convert, so if 7nm isn't far away it makes no sense to spend a ton of money on 10. That ship probably sailed 2 years ago. 10 will not be deployed outside the fab(s?) Intel used to develop it.

That is almost certainly why 8th gen 14nm parts are still around, causing youtube vloggers to freak the gently caress out about AHMAGAWD HOW COULD APPLE SHIP OLD AND BUSTED SKYLAKE DERIVATIVES IN THE BASE MACBOOK AIR 2020 MODEL?????!??!?!! It's not Apple (or Dell, or whomever, they're all doing it afaik), it's Intel positioning 10nm chips as premium products because they don't have the 10nm capacity to make it a mainstream option.

Cygni
Nov 12, 2005

raring to post



Does anybody have Strong Thoughts about any of the current BIOS?

I've personally used the latest ASUS, Asrock, and EVGA boards but I havent built on an MSI or Gigabyte in a while, and I hear they have caught up or passed the three I listed. Anybody have experience comparing?

Shaocaholica
Oct 29, 2002

Fig. 5E


I hate most consumer Winamp BIOSs. Dell has always had the cleanest BIOS UIs since as long as I can remember.

priznat
Jul 7, 2009

Let's get drunk and kiss each other all night.

We use a bunch of those hp mini PCs for labwork and they have a pretty nice BIOS, easy to PXE boot for Fog imaging etc.

Shaocaholica
Oct 29, 2002

Fig. 5E


priznat posted:

We use a bunch of those hp mini PCs for labwork and they have a pretty nice BIOS, easy to PXE boot for Fog imaging etc.

Screenshot? I'm curious. I could do this with my HP Z8 at work but its at work where I can't go.

Shaocaholica
Oct 29, 2002

Fig. 5E


Look at this from 14 years ago and this design probably goes back even further.



You've got a 1-deep tree menu on the left with a scroll bar.

Nice looking tables. Highlighted buttons. I'm counting at least 6 colors for the UI. I don't recall anything else coming close.

priznat
Jul 7, 2009

Let's get drunk and kiss each other all night.

Shaocaholica posted:

Screenshot? I'm curious. I could do this with my HP Z8 at work but its at work where I can't go.

Not at work either, would have to take a photo of it too.

It's pretty basic just a white screen with nicely done buttons and stuff.

My one beef is that it's not obvious whether you have "OK" or "cancel" selected, one is light blue and the other is light grey and it defaults to "cancel" being selected. Had a guy who is a bit colourblind and so had no idea which was which. Only way to tell is the "Ok" will ask for confirmation.

HalloKitty
Sep 30, 2005

Adjust the bass and let the Alpine blast


Shaocaholica posted:

Look at this from 14 years ago and this design probably goes back even further.



You've got a 1-deep tree menu on the left with a scroll bar.

Nice looking tables. Highlighted buttons. I'm counting at least 6 colors for the UI. I don't recall anything else coming close.

I agree.

Adbot
ADBOT LOVES YOU

D. Ebdrup
Mar 13, 2009



That's one thing I really don't like about SuperMicros non-server boards, they come with a lot of UEFI customization instead of sticking to the regular look that every other SuperMicro board has.

eames posted:

The buildzoid video on selective hyperthreading is quite interesting.

The idea, as he explains it, is that it allows you to pick up extra overclocking headroom by selectively disabling hyperthreading on the "weak" cores of a CPU. 10C/16T at 5.2 GHz will usually be faster than 10C/20T at 5.1 GHz, so disabling HT on the right cores can be a overall performance gain.
I guess it's a neat technical feature, but holy poo poo, 100MHz is what we're reduced to gaining?
You can barely show that on a microbenchmark with microarchitectural compile-time optimizations enabled, if you're benchmarking properly.

D. Ebdrup fucked around with this message at 18:36 on May 10, 2020

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply
«561 »