|
JazzmasterCurious posted:Oh poo poo yeah. Haven't had my morning coffee. Speaking of, thanks for the reminder, I should go grind some beans.
|
#
?
Jan 14, 2018 11:25
|
|
|
|
| # ? Apr 25, 2024 10:49 |
|
|
You can always hold off on upgrading your graphics card, but DDR4 is a disaster right now for sure. I'm hoping Samsung's v2 production will change things, but I have no idea how long it'd take for prices to be affected by a new supply.
|
|
#
?
Jan 14, 2018 13:20
|
|
|
ufarn posted:You can always hold off on upgrading your graphics card, but DDR4 is a disaster right now for sure. (unless the chinese companies that approached samsung pulled their strings and demanded a more drastic lowering of prices)
|
|
#
?
Jan 14, 2018 13:22
|
|
|
Wirth1000 posted:Death to cryptominers.
|
|
#
?
Jan 14, 2018 15:22
|
|
|
I'm glad I got my 5820k when I did after this thread inspired me to look at prices. Wowza! I don't see a reason to upgrade for years to come. Then again I only play slow strategy games and lock everything else to 30 FPS sooooo...
|
|
#
?
Jan 14, 2018 17:52
|
|
|
GRINDCORE MEGGIDO posted:Is there any reason to think the management engine will be optional, via bios setting, in newer chips?  Like they'd give up their spyware
|
|
#
?
Jan 14, 2018 17:56
|
|
|
Look at how cheap used Broadwell-Es are getting though. For anyone on X99 they seem like the "correct" upgrade option. For the cost of a Coffeelake system you could get a 6900K to drop into your existing hardware. They benchmark very favorably to Coffeelakes too, even in stuff that doesn't use 8 cores.
|
|
#
?
Jan 14, 2018 17:57
|
|
|
Azhais posted:
I'm thinking no as well.
|
|
#
?
Jan 14, 2018 19:34
|
|
|
Azhais posted:
Government mandated no doubt.
|
|
#
?
Jan 15, 2018 01:45
|
|
|
No, there's no mandate. It's just...extremely convenient...for...things...
|
|
#
?
Jan 15, 2018 01:46
|
|
|
Oddly Lenovo added the ability to lobotomize AMT out of my Broadwell detachable thing months ago.
|
|
#
?
Jan 15, 2018 01:57
|
|
|
Power saving and performance being number 1. For secure computers sure they should be able to disable it, but how many people want to go back to the performance issues and power use of a decade ago for an average computer?
|
|
#
?
Jan 15, 2018 01:59
|
|
|
Maybe power management being a separate module would be a logical thing they could do. Have people reported power management not being functional when using machines with disabled IME anyway? GRINDCORE MEGGIDO fucked around with this message at 02:19 on Jan 15, 2018 |
|
#
?
Jan 15, 2018 02:15
|
|
|
I suppose we're talking about standby and poo poo? Because I don't see a reason why the main CPU wouldn't be able to manage its own power during normal runtime?
|
|
#
?
Jan 15, 2018 02:42
|
|
|
The biggest feature I remember getting talked about from the ME is responsive throttling. Use 10 watts at idle and seamlessly ramp up to your 150 watt overclocked settings. No crashing or mistakes where it doesn't speed up. I think standby was around long before the ME.
|
|
#
?
Jan 15, 2018 02:49
|
|
|
craig588 posted:Look at how cheap used Broadwell-Es are getting though. For anyone on X99 they seem like the "correct" upgrade option. For the cost of a Coffeelake system you could get a 6900K to drop into your existing hardware. They benchmark very favorably to Coffeelakes too, even in stuff that doesn't use 8 cores. I'm still waiting on benchmarks for how hard X99 is going to be hit by the Spectre/Meltdown patch though, supposedly it's a bigger impact on older systems and it's not entirely clear where that dividing line is or just how bad the impact will be. But yeah, for gaming there's no solution quite like assloads of L3 cache and a ringbus, which is what X99 delivers in spades. The X99 chips are still parked on top for a lot of benchmarks, assuming you can get the clocks up to mid 4's. craig588 posted:Power saving and performance being number 1. For secure computers sure they should be able to disable it, but how many people want to go back to the performance issues and power use of a decade ago for an average computer? You can have power management without it needing DMA access to all of memory and a low-level tap on your network interfaces.
|
|
#
?
Jan 15, 2018 05:00
|
|
|
Paul MaudDib posted:You can have power management without it needing DMA access to all of memory and a low-level tap on your network interfaces. Now that's just madness.  Well you can't completly remove the ME anyways. I assume somewhere there do also lie the power management functions, I doubt they get knocked out if you lobotomize the redundant stuff, that would be an amazingly strange move as I'm sure certain entities did ask for the possibility to deactivate/remove most of the interesting functions without making the computer unusable. The important question that of course remains is: Can you run doom on it?
|
|
#
?
Jan 15, 2018 09:21
|
|
|
Police Automaton posted:
|
|
#
?
Jan 15, 2018 09:45
|
|
|
GRINDCORE MEGGIDO posted:Is there any reason to think the management engine will be optional, via bios setting, in newer chips? They'll probably segment out some of the capabilities of ME at some point but I wouldn't be holding my breath for it any time soon. That's a years long engineering undertaking to get out in to a product and the pressure is only starting to ramp up for real now. Azhais posted:
Shut up you loving moron.
|
|
#
?
Jan 15, 2018 17:09
|
|
|
Paul MaudDib posted:You can have power management without it needing DMA access to all of memory and a low-level tap on your network interfaces. By the account of Intel's own engineers, ME was a good idea that suffered from everyone bandwagoning on to it because a strong scope or security impact review was not enforced. OOBM stuff is really appealing to enterprise customers when actually implemented so of course they shipped that feature set (and you could turn it off back in the early iterations). But then you're presenting a stable, always on platform to target and stuff like power management got thrown in there as more of the chipset moved in to the CPU die despite the objections of some and now we have an interwoven mess that's going to take a long time to segregate the functions out of and phase out.
|
|
#
?
Jan 15, 2018 17:15
|
|
|
So we should all buy the 10770k or whatever after they’ve done a complete architecture redesign and the ME and branch prediction are properly separated out and things are better. That’ll be the next good Intel chip to buy.
|
|
#
?
Jan 15, 2018 18:13
|
|
|
I miss Cyrix. There I said it. Not ashamed.
|
|
#
?
Jan 15, 2018 18:20
|
|
|
This is a good ME talk where they stripped ME down to its minimal components to function without the chip killing the system. https://www.youtube.com/watch?v=wsmHmYxyoxg It's doable with a software patch but it's not something I would recommend doing with some random patch A Guy made. Intel needs to be pressured to restructure the internals of ME so potentially undesirable components like remote management can be disabled while preserving things like power management. Most likely they will not do anything for legacy chips and will use this type of effort on new hardware as a bulletpoint for refreshing hardware.
|
|
#
?
Jan 15, 2018 18:54
|
|
|
BangersInMyKnickers posted:Shut up you loving moron. A persistent back door which a normal end-user has zero control over. Yeah, that guy is totally dumb.
|
|
#
?
Jan 15, 2018 21:22
|
|
|
BangersInMyKnickers posted:This is a good ME talk where they stripped ME down to its minimal components to function without the chip killing the system. This is cool, thanks. I may have some BIOS source + ME firmware and toolkits from back in the Nehalem days, and I remember using a stripped-down ME image from Intel. Wonder if it’s time to use time off from work to gently caress around with doing custom neutered BIOSes for people. poo poo isn’t signed anyway, place on SPI flash and PCH will boot CPU from it. Packaging as delta patches against public BIOS binaries probably would sidestep IP issues.
|
|
#
?
Jan 16, 2018 00:53
|
|
|
If you could do the same stuff as the Pure(?) guys, and disable the ME, I would be interested. 
|
|
#
?
Jan 16, 2018 08:52
|
|
|
Cannonlake showing up on 3dmark: https://wccftech.com/intel-cannonlake-coffee-lake-desktop-mobility-cpus-leak/ Rest of the leaked part names for the March Coffee Lake spread were originally leaked back in November.
|
|
#
?
Jan 16, 2018 19:18
|
|
|
Cygni posted:Cannonlake showing up on 3dmark:
|
|
#
?
Jan 16, 2018 19:33
|
|
|
Mr Shiny Pants posted:If you could do the same stuff as the Pure(?) guys, and disable the ME, I would be interested. Looks like Purism? I can't tell from their website if they are actually selling laptops with a fully functional coreboot EFI setup (or if it's just the 13" 2017 models), or if that's a WIP. Are they still doing binary blobs for Intel's MRC and the other black boxes that are needed?
|
|
#
?
Jan 16, 2018 21:49
|
|
|
I apparently killed a PCIe slot on one of my X99 boards a while back, Gigabyte are being bitches about RMA'ing it despite it being less than a year since purchase. I've been putting off doing anything about it because I was scared of bending pins, finally sacked up today and swapped it over and oh my god LGA2011 is so much easier to work with than LGA1151  The lip on LGA1151 sockets go up around all four sides of the chip, LGA2011 only has the socket going around 2 sides, which actually make it possible for my fat fingers to grab onto the sides and just lift it out of the socket without awkwardly trying to grab onto 1/4" of recessed heatspreader like I have to do with LGA1151. I'm irrationally terrified of it slipping halfway out, slapping onto the pins, and  motherboard.Intel apparently makes a "field installation/removal" tool for their various sockets (including one for LGA115x) that clips onto the heatspreader, I'm convinced that I need to buy one, but I can't find it anywhere. Paul MaudDib fucked around with this message at 09:27 on Jan 17, 2018 |
|
#
?
Jan 17, 2018 09:19
|
|
|
Does an undervolt through XTU require the program to stay open and/or when booted?
|
|
#
?
Jan 17, 2018 09:46
|
|
|
I think there's a service that runs in the background but you don't need to actually open the UI.
|
|
#
?
Jan 17, 2018 10:45
|
|
|
The XTU service is supposed to run on startup automatically but I read that sometimes wouldn't so I setup a scheduled task on my laptop to run the service on login if it wasn't already running. You can google around and find out the steps for that but it's pretty straightforward.
|
|
#
?
Jan 17, 2018 14:12
|
|
|
movax posted:Looks like Purism? I can't tell from their website if they are actually selling laptops with a fully functional coreboot EFI setup (or if it's just the 13" 2017 models), or if that's a WIP. Are they still doing binary blobs for Intel's MRC and the other black boxes that are needed? Purism, that's it. I don't know how they do it, but they sell a laptop which is fully functional and has the ME disabled.
|
|
#
?
Jan 17, 2018 16:44
|
|
|
Paul MaudDib posted:I apparently killed a PCIe slot on one of my X99 boards a while back, Gigabyte are being bitches about RMA'ing it despite it being less than a year since purchase. I've been putting off doing anything about it because I was scared of bending pins, finally sacked up today and swapped it over and oh my god LGA2011 is so much easier to work with than LGA1151 I'm sure you know this but if the CPU isn't making good contact with all the LGA pins you can get stuff like PCIe slots not working. Definitely worth checking. [edit] a couple times I've bent my mobo putting in memory and gfx cards enough to make the CPU lose contact with the pins. I think I finally fixed it by removing everything and reassembling outside the case, also loosening the heatsink a bit (aftermarket adjustable one). redeyes fucked around with this message at 01:19 on Jan 18, 2018 |
|
#
?
Jan 17, 2018 23:12
|
|
|
Rastor posted:Intel hosed up the microcode update for Haswell (Core i3/5/7 4000 series) chips for desktop and mobile and Broadwell (Core i3/5/7 5000 series) chips for mobile BTW it turns out the microcode update is also causing crashes on Ivy Bridge, Sandy Bridge, Skylake, and Kaby Lake https://www.theverge.com/2018/1/18/16904726/intel-meltdown-spectre-fix-higher-reboots
|
|
#
?
Jan 18, 2018 16:35
|
|
|
Finally installed the latest BIOS for my ASUS Z77 motherboard. From September of 2013. Also not installing update that actually is available for my HTPC H97 board because stability is the one critical thing I need and rebooting all the time would ruin it's purpose. Thanks, companies. Well done.
|
|
#
?
Jan 18, 2018 19:09
|
|
|
I've got an i5 4790k in my system. I had been debating upgrading to whatever the newest architecture is, maybe even jump from i5->i7 to make stuff like EU4/CK2/etc run faster but I'm getting the sense now that I should just wait till late 2018/early 2019 to consider any upgrades to the CPU/Motherboard. Is that accurate? edit: Also it seems like I should rely on just the Windows update and not try and flash my BIOS or anything to fix the Meltdown/Spectre bug. That correct? axeil fucked around with this message at 22:01 on Jan 18, 2018 |
|
#
?
Jan 18, 2018 21:52
|
|
|
axeil posted:I've got an i5 4790k in my system. I had been debating upgrading to whatever the newest architecture is, maybe even jump from i5->i7 to make stuff like EU4/CK2/etc run faster but I'm getting the sense now that I should just wait till late 2018/early 2019 to consider any upgrades to the CPU/Motherboard. BIOS may be needed for Spectre, Meltdown should be fixed by Windows update.
|
|
#
?
Jan 18, 2018 22:58
|
|
|
|
| # ? Apr 25, 2024 10:49 |
|
|
Spectre is also partially mitigated with recent browsers by isolating tabs to their own process. But other individual apps would need to be restricted similarly.
|
|
#
?
Jan 18, 2018 23:05
|
|