|
mewse posted:Coffee lake (8000 series) boosted core counts to compete with Ryzen So what’s the sweet spot price/performance intel core count now? 8C/16T?
|
# ? Mar 20, 2018 18:51 |
|
|
# ? Apr 18, 2024 08:37 |
|
Shaocaholica posted:So what’s the sweet spot price/performance intel core count now? 8C/16T? The i7-8700k is generally recognized as the best gaming processor and its 6c/12t
|
# ? Mar 20, 2018 18:55 |
|
Are modern games even taking advantage of that or is it more to cover the overhead of capturing/steaming?
|
# ? Mar 20, 2018 18:59 |
|
Shaocaholica posted:Are modern games even taking advantage of that or is it more to cover the overhead of capturing/steaming? They have enough threads to put one or more on each core, though they generally do get bottlenecked by the mainthread to some extent.
|
# ? Mar 20, 2018 19:00 |
|
mewse posted:The i7-8700k is generally recognized as the best gaming processor and its 6c/12t 8700k is probably not the price/perf sweet spot though. The 8600 at 6c/6t is probably a better use of money if you are constrained to a budget, 8400 even a step better. Right now its rare for games to be that limited by core count or even raw speed.
|
# ? Mar 20, 2018 19:03 |
|
Shaocaholica posted:Are modern games even taking advantage of that or is it more to cover the overhead of capturing/steaming? If you only open a game and have nothing else open then it matters a whole lot less.
|
# ? Mar 20, 2018 19:13 |
|
Shaocaholica posted:Are modern games even taking advantage of that or is it more to cover the overhead of capturing/steaming?
|
# ? Mar 20, 2018 19:15 |
|
I'm hoping a fool's hope that we see an 8c/8t "Core i5-9400" for ~$250 by the end of the year.
spasticColon fucked around with this message at 20:05 on Mar 20, 2018 |
# ? Mar 20, 2018 20:02 |
|
spasticColon posted:I'm hoping a fool's hope that we see an 8c/8t "Core i5-9400" for ~$250 by the end of the year.
|
# ? Mar 20, 2018 20:29 |
|
I think Intel would rather bring hyper-threading to the i5 (for the love of god already loving please) than put actual cores on deep discount.
|
# ? Mar 20, 2018 20:34 |
|
Lol wut there no mainstream 4c/8t model anymore? E: oh I guess that’s the 6c/12t now
|
# ? Mar 20, 2018 21:00 |
|
Craptacular! posted:I think Intel would rather bring hyper-threading to the i5 (for the love of god already loving please) than put actual cores on deep discount. But that would make the i7 chips obsolete/irrelevant unless you mean for the CL 9xxx series they make the i5-9xxx chips 6c/12t and make the i7-9xxx chips 8c/16t. That would actually make more sense. spasticColon fucked around with this message at 21:50 on Mar 20, 2018 |
# ? Mar 20, 2018 21:47 |
|
Cygni posted:Intel has now come out and said that Cascade Lake (the next Xeon/HEDT platform) and the new desktop product refresh due at the end of this year (Cannon Lake? Ice Lake?) will be hardware immune to Meltdown and Spectre v2. Question; are the coded fixes equal to the innate hardware fix that's coming?
|
# ? Mar 22, 2018 03:30 |
|
Pastry Mistakes posted:Question; are the coded fixes equal to the innate hardware fix that's coming? Nobody knows. 12-18 months is a pretty quick turnaround for a bug that severe; on the other hand it's not like they have to solve this from the ground up. AMD is much less affected so all they really need to do is rip off what AMD's doing. Which means not sharing Branch Target Buffer aliases between multiple addresses, and doing security checks before issuing the speculative execution rather than after it completes. I'd say those are not trivial tasks, but they should be relatively discrete tasks, there is no need to "redesign the whole core" or anything like that.
|
# ? Mar 22, 2018 04:03 |
|
Well, I can confirm that a BIOS update from Intel is available that patches the Xeon E3-1270v6 and it works fine. Confirmed the vulnerabilities are patched via the Microsoft PowerShell script that checks the CPU. I have yet to notice a performance degradation.
|
# ? Mar 23, 2018 03:30 |
|
Interesting sorta post mortem on the Spectre/Meltdown patches for Intel. As with the other testing, shows pretty much no impact to gaming numbers and anywhere from "unnoticeable" to "goddamnit" impacts to storage performance depending on what you are doin. https://www.anandtech.com/show/12566/analyzing-meltdown-spectre-perf-impact-on-intel-nuc7i7bnh
|
# ? Mar 23, 2018 22:26 |
|
Cygni posted:Interesting sorta post mortem on the Spectre/Meltdown patches for Intel. As with the other testing, shows pretty much no impact to gaming numbers and anywhere from "unnoticeable" to "goddamnit" impacts to storage performance depending on what you are doin. I think there is a chance for the storage impact to be somewhat mitigated. It only seems to affect NVMe drives under Windows, and it might be possible to change how the NVMe driver behaves to help out. Hopefully this can be explored by MS (and Samsung, who like to write their own driver) and an improvement found.
|
# ? Mar 24, 2018 01:36 |
|
EoRaptor posted:I think there is a chance for the storage impact to be somewhat mitigated. It only seems to affect NVMe drives under Windows, and it might be possible to change how the NVMe driver behaves to help out. Hopefully this can be explored by MS (and Samsung, who like to write their own driver) and an improvement found. It might not be possible to fix this with a driver patch. Meltdown mitigation adds overhead to all transitions between kernel mode and user mode. Generally speaking, one IOP equals one kernel/user round trip, so it's likely that the cause of the slowdown is just that a NVME disk and driver are capable of cranking out so many IOPs that the additional system call overhead really hurts. (that's why the article mentions SATA probably doesn't notice this as much -- it's just that SATA can't sustain as many IOPs.)
|
# ? Mar 24, 2018 07:49 |
|
March 2018: Intel discovers multicore optimization for game engines. Render need better jobify. https://www.computerbase.de/2018-03/multi-core-cpu-games/
|
# ? Mar 27, 2018 23:09 |
|
eames posted:March 2018: Intel discovers multicore optimization for game engines. Render need better jobify. mystes fucked around with this message at 23:41 on Mar 27, 2018 |
# ? Mar 27, 2018 23:38 |
|
eames posted:March 2018: Intel discovers multicore optimization for game engines. Render need better jobify.
|
# ? Mar 27, 2018 23:56 |
|
mystes posted:Gee, I wonder whose fault it is that games don't bother doing a good job of optimizing for more than 4 cores. Maybe the company that decided not to bother going beyond 4 cores in desktop processors for 10 years when they didn't face much competition? Well it allowed me to game on a 2500K for 7 years so I'm not going to complain too much.
|
# ? Mar 28, 2018 00:56 |
|
mystes posted:Gee, I wonder whose fault it is that games don't bother doing a good job of optimizing for more than 4 cores Game developers. 6 thread 360s were out in 2005 and the 2 thread + 6 core heterogeneous core design PS3s were out in 2006. PC people melted the gently caress down over it when a developer finally brought heavy core usage over in the form of GTA IV, which refused to play nicely with everybody's old 2 core systems.
|
# ? Mar 28, 2018 01:58 |
|
https://www.techspot.com/news/73896-branchscope-attack-successfully-demonstrated-several-intel-cpus.html https://arstechnica.com/gadgets/2018/03/its-not-just-spectre-researchers-reveal-more-branch-prediction-attacks/ Another branch predictor exploit similar to Spectre has been demonstrated on several Intel CPUs.
|
# ? Mar 28, 2018 05:49 |
|
OhFunny posted:https://www.techspot.com/news/73896-branchscope-attack-successfully-demonstrated-several-intel-cpus.html From the article: quote:Update: in a statement, Intel says: We have been working with these researchers and we have determined the method they describe is similar to previously known side channel exploits. We anticipate that existing software mitigations for previously known side channel exploits, such as the use of side channel resistant cryptography, will be similarly effective against the method described in this paper. Could it be that these are already patched?
|
# ? Mar 28, 2018 06:02 |
|
fishmech posted:Game developers. 6 thread 360s were out in 2005 and the 2 thread + 6 core heterogeneous core design PS3s were out in 2006. Wasn't one of the cores/threads reserved for system UI so the game wasn't allowed to touch them? And you're still severely limited by memory constraints so there's only so much you can do with one or two more threads anyways. I'm not a console game developer so I don't know all the details, but I find it hard to believe that the guys who had been desperately squeezing every last inch of performance out of that hardware for years would leave extra threads unexamined.
|
# ? Mar 28, 2018 09:14 |
|
isndl posted:Wasn't one of the cores/threads reserved for system UI so the game wasn't allowed to touch them? And you're still severely limited by memory constraints so there's only so much you can do with one or two more threads anyways. Yep, add to that that the PS3 was something else altogether with its SPUs.
|
# ? Mar 28, 2018 10:05 |
|
OhFunny posted:https://www.techspot.com/news/73896-branchscope-attack-successfully-demonstrated-several-intel-cpus.html Paging Paul.
|
# ? Mar 28, 2018 14:44 |
|
fishmech posted:Game developers. 6 thread 360s were out in 2005 and the 2 thread + 6 core heterogeneous core design PS3s were out in 2006. tbf that game melted down on my contemporaneous 4c/4t system as well; the performance metrics were really variable on similar systems, it didn't make much sense
|
# ? Mar 28, 2018 15:32 |
|
PerrineClostermann posted:Paging Paul. yet another motherfucking branch attack boy I hope Cascade Lake actually has the thing (it probably won't)
|
# ? Mar 28, 2018 15:38 |
|
Since we're talking about Meltdown... Microsoft's Windows 7 Meltdown fixes from January, February made PCs MORE INSECURE
|
# ? Mar 28, 2018 21:41 |
|
microsoft still being in business is proof that the free market doesnt work
|
# ? Mar 28, 2018 21:46 |
|
Lol Windows 7. End of Extended Support can't come soon enough.
|
# ? Mar 28, 2018 22:15 |
|
Windows 7 is probably gonna hang on as long as XP.
|
# ? Mar 28, 2018 23:13 |
|
I'm sure glad MS dropped that QA team. We've seen zero fallout from that decision.
|
# ? Mar 28, 2018 23:21 |
|
isndl posted:Wasn't one of the cores/threads reserved for system UI so the game wasn't allowed to touch them? And you're still severely limited by memory constraints so there's only so much you can do with one or two more threads anyways. PS3 had 8 of the heterogenous cores of which 1 was always off due to yield optimization, and 1 was OS reserved. 360 had one thread reserved. Total threads available for the games: 5 powerpc threads on 3 cores for the 360, 2 powerpc threads on one core and 6 "spe" cores on the PS3. Very few games bothered to use nearly the same level of threading on the PC until well into the console lives, again with GTA IV's pc port being one of the first.
|
# ? Mar 28, 2018 23:55 |
|
One of the probably SDK supported optimizations considering how many games used it on the PS3 was letting the cores race and the first one to finish would be the result used. They discovered that in developing RPCS3, pausing threads would result in no difference but a large performance improvement on the PC since it wasn't racing all the cores.
|
# ? Mar 29, 2018 00:12 |
|
Generic Monk posted:tbf that game melted down on my contemporaneous 4c/4t system as well; the performance metrics were really variable on similar systems, it didn't make much sense Yeah, it was hell on GPU’s as well, you’d better have 1 gig of VRAM or you were going to have a bad time, and only brand-new cards at the time had that much memory. You could run it with 512 but the pop-in was so bad that some parts of the game were nearly unplayable.
|
# ? Mar 29, 2018 05:07 |
|
Generic Monk posted:tbf that game melted down on my contemporaneous 4c/4t system as well; the performance metrics were really variable on similar systems, it didn't make much sense GTA 4 has some absolutely nutty exponential load scaling with resolution, meaning that 10 years from now it will still run like rear end at 8K. It is to GPU usage what Oblivion is to single core performance.
|
# ? Mar 29, 2018 05:19 |
|
|
# ? Apr 18, 2024 08:37 |
|
Saints row always ran pretty well, another reason the series is clearly superior.
|
# ? Mar 29, 2018 18:43 |