|
Everything just works! *Shoulders military rucksack full of dongles, because ports aren't pretty*
GRINDCORE MEGGIDO fucked around with this message at 17:40 on Jul 18, 2018 |
#
?
Jul 18, 2018 17:37
|
|
|
|
| # ? May 24, 2024 16:41 |
|
|
eames posted:My point is that this shows that TDP ratings seem to be inaccurate. Where are those numbers from?
|
|
#
?
Jul 18, 2018 17:59
|
|
|
fishmech posted:Macs aren't top end laptops except in price though. What on earth does that mean?
|
|
#
?
Jul 18, 2018 18:54
|
|
|
mewse posted:The problem is if you buy the new i9 and try to run video encoding you will get worse job times than the same model with an i7 skylake from last year That could be caused by any number of issues though, most of which are going to be Apple's fault, whether it be because of how they implemented cooling or macOS because he ran it in macOS for some weird reason. If I remember right other high performance laptops have throttled because of a lack of VRM cooling rather than the problem being cooling the CPU itself. Winks fucked around with this message at 19:27 on Jul 18, 2018 |
|
#
?
Jul 18, 2018 19:19
|
|
|
Dr. Fishopolis posted:What on earth does that mean? What part do you not understand?
|
|
#
?
Jul 18, 2018 19:24
|
|
|
Paul MaudDib posted:Where are those numbers from? 2018 is from a screenshot on macrumors, 2015 is my own. I can post links later if you want, currently on phone.
|
|
#
?
Jul 18, 2018 19:31
|
|
|
fishmech posted:What part do you not understand? He's confused because you are wrong. They are very much high end laptops by multiple measures. It doesn't mean that it isn't some lovely design if they truly are overheating that easily, but these things are NICE.
|
|
#
?
Jul 18, 2018 20:22
|
|
|
LRADIKAL posted:He's confused because you are wrong. They are very much high end laptops by multiple measures. It doesn't mean that it isn't some lovely design if they truly are overheating that easily, but these things are NICE. No, they're not. They're just expensive crap. They don't compare. Apple has refrained from doing high end equipment for years in favor of their thinness obsession. This latest overheating example is just one more to toss on the pile. Or the way their "high end" desktop was a dumb tube with minimal updates for 5 years and so on.
|
|
#
?
Jul 18, 2018 20:42
|
|
|
Don't forget the keyboards that are riveted in so you can't replace them if they break Also a single grain of sand breaks them
|
|
#
?
Jul 18, 2018 21:45
|
|
|
Today is Intel's 50th anniversary 
|
|
#
?
Jul 18, 2018 21:54
|
|
|
canyoneer posted:Today is Intel's 50th anniversary Happy Birthday e: I got a nice little metal plaque on the 40th anniversary, with what appeared to be a 2.5 core Nehalem-based design drawn on one side
|
|
#
?
Jul 18, 2018 22:13
|
|
|
JawnV6 posted:Happy Birthday I have that buried in my desk drawer somewhere... e: My Merom one is the only one hanging in my cube because it's the only one they put velcro on the back
|
|
#
?
Jul 18, 2018 22:24
|
|
|
I want to see someone crack open one of the new i9 MacBook Pros and shove some liquid metal in there. Even if it's just a bare shim. DO IT. der8auer don't let me down
|
|
#
?
Jul 19, 2018 01:22
|
|
|
Arivia posted:I want to see someone crack open one of the new i9 MacBook Pros and shove some liquid metal in there. Even if it's just a bare shim. DO IT. der8auer don't let me down I'm pretty sure I saw a teardown where they opened a recent macbook pro and a bunch of glitter and unicorn blood poured out and they were never able to reassemble it
|
|
#
?
Jul 19, 2018 01:27
|
|
|
Computer archaeology question time: I'm futzing around with old server hardware. Specifically, x3450 xeon + supermicro X8SI6-F board. I have no idea what kind of RAM this thing can take, because it makes no sense to me. Apparently even though it takes registered ECC 1333 it can only really run at DDR3-800 if you put anything more than 8GB in for reasons? And there's 16GB DDR3 but it caps out at 8gb because an extra pin for the address lines was too much to ask for. And single, dual and quad rank dimms, and separate memory limitations (and speed limitations) for each. The gently caress was everyone smoking back then? http://ftpw.supermicro.com.tw/products/Memory/3400_support.cfm?pname=MBD-X8SI6-F&DIMM=6 E: Apparently Intel memory controllers were bad and until ivy-e were super weird about what RAM they would take. AMD stuff just worked for all the standard sizes. What I can't figure out is if it supports dual-rank 8gb modules, (4gb/rank) or if it only supports quad for 8gb. quad rank runs at half the speed. E: Nope. I forgot how bad computers were back then. 16gb of ddr3-1333 or 24,32 at -800. That's just awful. Harik fucked around with this message at 14:11 on Jul 19, 2018 |
|
#
?
Jul 19, 2018 11:51
|
|
|
https://www.digitimes.com/news/a20180614PD212.htmlquote:Samsung Electronics, SK Hynix and Micron Technology - the world's top-3 DRAM vendors - could be fined up to US$8 billion in China over alleged price fixing, according to industry sources.
|
|
#
?
Jul 23, 2018 02:38
|
|
|
Be nice if we could get some money back after being ripped off for the last 3 years. Not going to happen
|
|
#
?
Jul 23, 2018 02:58
|
|
|
I just hope this results in ram prices going down again.
|
|
#
?
Jul 23, 2018 21:13
|
|
|
Rumor's going around that Intel's disabling hypethreading on the i7 going forward with the 9xxx series and making it an i9 feature only.
|
|
#
?
Jul 25, 2018 16:53
|
|
|
Combat Pretzel posted:Rumor's going around that Intel's disabling hypethreading on the i7 going forward with the 9xxx series and making it an i9 feature only. Didn't they just add hyperthreading to laptop i5 chips
|
|
#
?
Jul 25, 2018 17:24
|
|
|
I don't know then. The i7-9700K is rumored to have 8C/8T "only", and I'm not sure why they'd do that. Does using hyperthreading (or not) make a huge difference in temperature? That CPU and the 9900K are also rumored to have very high boost clocks. --edit: Apparently the 9900K has hyperthreading.
|
|
#
?
Jul 25, 2018 17:41
|
|
|
Combat Pretzel posted:Rumor's going around that Intel's disabling hypethreading on the i7 going forward with the 9xxx series and making it an i9 feature only. I'm not sure why they're doing this (market segmentation? binning? power consumption?) but it's fine with me. I'm pretty sure if somebody turned off HT in my BIOS I'd never notice a difference. Those who are affected by this change (video encoding, 3D rendering, multiple VMs) are probably better off with AMD anyway.  fake edit: yes, toggling HT shows a significant difference in power consumption and temperature in applications that make use of it, though performance is also affected. 8700Ks with HT turned off also tend to overclock 100-200 MHz higher. fake edit 2: maybe they're releasing the 8/8 with "sane" power consumption with Z370 compatibility and are saving the 8/16 for Z390 with beefier VRMs. I'd be fine with that. eames fucked around with this message at 17:45 on Jul 25, 2018 |
|
#
?
Jul 25, 2018 17:43
|
|
|
If the rumour is accurate they may have decided it's a good idea to trash hyperthreading in order to boost single core performance and also to head off some security vulnerabilities that are coming down the pipe (openbsd has disabled hyperthreading by default)
|
|
#
?
Jul 25, 2018 17:44
|
|
|
Conceptually HT is loading two threads in to a processing unit simultaneously. At various points one of the theads will be pulling instructions off the register or something that doesn't die up the execution unit so it lets that other thread utilize the resources of the core that would have otherwise been idle. This can dramatically improve performance on certain workloads but yeah, its going to up your TDP because you're lighting up more components of your processor for more clock cycles when they would have otherwise been idle.
|
|
#
?
Jul 25, 2018 17:52
|
|
|
Computerbase posted 9900K/9700K/9600K specs from a reasonably reliable source, they seem to match the recently leaked benchmark. https://www.computerbase.de/2018-07/spezifikationen-core-i9-9900k-i7-9700k-i5-9600k/ (95W TDP with 8x4.7GHz allcore turbo, ha)
|
|
#
?
Jul 25, 2018 18:14
|
|
|
BangersInMyKnickers posted:Conceptually HT is loading two threads in to a processing unit simultaneously. At various points one of the theads will be pulling instructions off the register or something that doesn't die up the execution unit so it lets that other thread utilize the resources of the core that would have otherwise been idle. This can dramatically improve performance on certain workloads but yeah, its going to up your TDP because you're lighting up more components of your processor for more clock cycles when they would have otherwise been idle. Moreover, I'm pretty sure that in order to serve two threads at once the processor is going to have to light up two decoder units per core instead of one - this will increase power usage by some amount regardless of whether the second thread is actually able to use any of the execution units at any given time, although probably not by very much by itself. Eletriarnation fucked around with this message at 18:30 on Jul 25, 2018 |
|
#
?
Jul 25, 2018 18:20
|
|
|
eames posted:(95W TDP with 8x4.7GHz allcore turbo, ha) AMD really seems to have them running scared, since they're stuck on 14nm
|
|
#
?
Jul 25, 2018 18:20
|
|
|
Eletriarnation posted:Moreover, I'm pretty sure that in order to serve two threads at once the processor is going to have to light up two decoder units per core - this will increase power usage by some amount regardless of whether the second thread is actually able to use any of the execution units at any given time, although probably not by very much by itself. There's a temptation to only cover a minimal set. Huge things that both threads could potentially use, such as the caches, wouldn't have to be tagged for performance or correctness. But in the era of prefetches running everything, you'd only want a cache hit on something that thread brought in.
|
|
#
?
Jul 25, 2018 18:24
|
|
|
JawnV6 posted:What? You think there's an extra decoder just lyin' around in there in case HT is enabled? HT is tagging some structures with an extra 'thread' bit. Like every uop going down the pipeline is associated with thread 0 or 1. You don't literally duplicate every functional block and drop a bunch of spares around. I did, but my experience on processor design only goes as far as undergraduate classes - we didn't get into superscalar design, let alone SMT. Thanks for the correction.
|
|
#
?
Jul 25, 2018 18:31
|
|
|
The i9 branding is bumping i7 down the stack, basically.
|
|
#
?
Jul 25, 2018 19:19
|
|
|
likely with a substantial price increase for the top end desktop SKU without the outrage because it is an entirely new product
|
|
#
?
Jul 25, 2018 19:24
|
|
|
Is it possible that disabling HT is for a specific series only? Disabling HT to spine single core performance would probably make a good CPU for gaming.
|
|
#
?
Jul 25, 2018 19:36
|
|
|
Golem.de claims the i9 will be soldered: https://www.golem.de/news/core-i9-9900k-intels-achtkerner-ist-verloetet-1807-135685.html
|
|
#
?
Jul 25, 2018 19:52
|
|
|
Dr. Fishopolis posted:What on earth does that mean? It's a Fischmech post
|
|
#
?
Jul 25, 2018 20:16
|
|
|
Any rough idea on how the 8C/8T i7 would compare to the current 6C/12T i7? I.e., does the addition of two cores even out the loss of four threads?
|
|
#
?
Jul 25, 2018 20:18
|
|
|
Space Racist posted:Any rough idea on how the 8C/8T i7 would compare to the current 6C/12T i7? I.e., does the addition of two cores even out the loss of four threads?  Napkin math This benchmark has the i5-8600 at ~13k and the i7-8700 at ~15k so the extra 6 threads are clearly not doubling performance. If we assume performance will scale linearly, we take the 6 core performance of the i5-8600, divide by 6 and multiply by 8 for 17104. IMO 2 additional real cores are going to give clear performance benefit
|
|
#
?
Jul 25, 2018 20:31
|
|
|
Space Racist posted:Any rough idea on how the 8C/8T i7 would compare to the current 6C/12T i7? I.e., does the addition of two cores even out the loss of four threads? Generally, at best a hyperthread is another 30% of a real core. In gaming, more like 10% impact on averages, although the impact is greater on minimums. AMD gets a bit more out of their SMT implementation, it's more like 50% at best. Of course, it's running on the same execution units, so if you are really nailing 100% load on some task, the hyperthread can also not help much at all, or even hurt (although this is rare these days). I suspect the 8C8T will have faster averages and similar minimums, or possibly very slightly worse minimums in some edge-case scenarios, compared to a 6C12T. It's still a step up.
|
|
#
?
Jul 25, 2018 20:55
|
|
|
I believe the 8C/8T with a soldered IHS, no HT and some process refinements should overclock to at least 5.0GHz if not more. At that point it should easily outperform a OCed 8700K unless the application has unusually good HT scaling. The 9700K should even run cooler than a delidded 8700K because it will dissipate roughly the same amount of heat over a larger die area. Readjusting the lineup is a smart move, the target audience won’t make use of hyperthreading anyway, not with 8 cores. The 8600K was different because one could argue that 6C/6T won’t be enough to run game engines built for next gen consoles.
|
|
#
?
Jul 25, 2018 22:39
|
|
|
My experience with i7s with HT is they are much more responsive under load. I'm going to miss that unless somehow having more cores makes it a non issue.
|
|
#
?
Jul 25, 2018 23:38
|
|
|
|
| # ? May 24, 2024 16:41 |
|
|
redeyes posted:My experience with i7s with HT is they are much more responsive under load. I'm going to miss that unless somehow having more cores makes it a non issue. Yeah, tbh the real benefit of HT is that you can squeeze in some background tasks during gaps in the main workload without incurring the cost of a context switch off your game/whatever. The measurable impact for games is minimum frametimes, for other stuff it's responsiveness. HT is not really worth it by itself as far as average framerate is concerned. It helps some but it's definitely not the same as a real core, especially in productivity workloads that are blasting one set of execution units really hard. Paul MaudDib fucked around with this message at 00:22 on Jul 26, 2018 |
|
#
?
Jul 26, 2018 00:19
|
|