|
evilweasel posted:Because they went with AMD processors, which were cheaper, and figured that they could force people to code for eight cores to make up for the lovely per-core power because at least with consoles you've got a fixed hardware spec. That's not quite the whole story, because they're Jaguar cores, netbook class. Nowhere near as good as even Bulldozer. But, they are low power, which I would guess is the reason they were chosen. They are cheap cores that don't chow down on much power. Ideal for use in an inexpensive small box when you don't want to make too much noise. Throwing 8 of them in just means they could use some for the OS, and leave the multi-threading of games up to developers.
|
# ? Feb 17, 2017 16:11 |
|
|
# ? Jun 10, 2024 11:01 |
|
Anime Schoolgirl posted:2000-2001 was also a time when Intel started using every trick (both legal and illegal) possible to make up for getting owned by Jim Keller's second brainchild and managed to hit a two-decade jackpot in the form of anticompetitive laptop contracts. Josh Lyman posted:Intel was getting killed in those P3 days though.
|
# ? Feb 17, 2017 16:12 |
|
For what it's worth (nothing?), I just ran the Blender benchmark on a dual Xeon X5680 3.33ghz / 48GB workstation (2011 Westmere) . Getting a time in the 27s, so 2s slower than the Ryzen machine with 50% extra cores but 0.12ghz deficit. (Disclaimer: I was using Blender v2.71 not 2.78a) 32.5s if I limit to a matching 16 threads.
Pablo Bluth fucked around with this message at 16:22 on Feb 17, 2017 |
# ? Feb 17, 2017 16:14 |
|
It's way more likely that Intel didn't have the time to design a proper GPU core and saw it as quite low margin for the amount of effort they'd have to put in. Not impossible for Intel to make a decent GPU, just not worth their time.
|
# ? Feb 17, 2017 16:19 |
|
Pablo Bluth posted:For what it's worth (nothing?), I just ran the Blender benchmark on a dual Xeon X5680 3.33ghz / 48GB workstation (2011 Westmere) . Getting a time in the 27s, so 2s slower than the Ryzen machine with 50% extra cores but 0.12ghz deficit. (Disclaimer: I was using Blender v2.71 not 2.78a) So that would be what, 8 core 2 CPU Intel vs 8 core 1 CPU Ryzen? Yeah.. a wash.
|
# ? Feb 17, 2017 16:20 |
|
The X5680 is a 6c/12t.
|
# ? Feb 17, 2017 16:24 |
|
Pablo Bluth posted:The X5680 is a 6c/12t. Oh so 12 core vs 8 core? Yeah thats hard to form any kind of opinion about.
|
# ? Feb 17, 2017 16:28 |
|
Well, that, and it's from 2008/9.
|
# ? Feb 17, 2017 16:30 |
|
Speaking of AMDs Blender benchmark, the new version of Blender snuck this inquote:Multiple improvements for the latest AVX2 CPUS: That's unfortunate timing for AMD, any reviewers who replicate their test will probably get more favorable results on Intel chips now
|
# ? Feb 17, 2017 16:36 |
What happens if you change the processor affinity for the blender benchmark to match 8c16t?
|
|
# ? Feb 17, 2017 16:37 |
|
repiv posted:That's unfortunate timing for AMD, any reviewers who replicate their test will probably get more favorable results on Intel chips now
|
# ? Feb 17, 2017 16:58 |
|
Watermelon Daiquiri posted:What happens if you change the processor affinity for the blender benchmark to match 8c16t? I think the conclusion is it's definitely time to request some new shinier workstations.
|
# ? Feb 17, 2017 17:01 |
|
Boiled Water posted:This leaves the question why pay for eight lovely cores when four would've done the job? Because 4 cores wouldn't be available fast enough to do what they wanted, at a price the console makers could stand while keeping the console affordable. And don't forget that the Xbox 360 was already 3 core/6 thread. So many common game engines already had optimization for at least that many threads. This sometimes even caused problems for PC ports: games like GTA IV on the PC just plain refused to run well on single core or dual core systems that people had at the time the ports came out, but ran great on higher core/thread counts afterwards.
|
# ? Feb 17, 2017 17:02 |
|
Wild speculation 1888 @ 3.5
|
# ? Feb 17, 2017 18:29 |
|
PC LOAD LETTER posted:That is certainly possible but IIRC there were rumors that Intel did a surprisingly low bid for the original Xbox CPU. I don't think the legit CPU cost was ever leaked but I do remember some public comments that AMD thought they had that contract sewed up and were shocked by how low Intel was willing to go. I thought Intel's deal let them charge the same price throughout the life of the console. So instead of the cost going down over the life of the console, Intel kept getting to charge the same price for a laughably out-dated CPU
|
# ? Feb 17, 2017 18:39 |
|
Risky Bisquick posted:Wild speculation is that good or bad?
|
# ? Feb 17, 2017 18:43 |
|
wargames posted:is that good or bad? That's better than what I'm seeing from a 4.5GHz 6700k so I'd say really, really good. I have no idea what that benchmark consists of though.
|
# ? Feb 17, 2017 18:44 |
|
WhyteRyce posted:I thought Intel's deal let them charge the same price throughout the life of the console. So instead of the cost going down over the life of the console, Intel kept getting to charge the same price for a laughably out-dated CPU Its possible they would've eventually made money somehow but the deal seemed to be more about trying to keep AMD from getting a profitable niche, even a low profit one (AMD supposedly was pitching a cheap Duron 800-900Mhz variant), rather than making money. wargames posted:is that good or bad?
|
# ? Feb 17, 2017 18:47 |
|
wargames posted:is that good or bad? Literally anyone with Windows can try that out for themselves for free in moments, so
|
# ? Feb 17, 2017 21:10 |
|
Anime Schoolgirl posted:2000-2001 was also a time when Intel started using every trick (both legal and illegal) possible to make up for getting owned by Jim Keller's second brainchild and managed to hit a two-decade jackpot in the form of anticompetitive laptop contracts. 2000-2001 was when Netburst was superseding the PIII but Intel had a lot of fab space available for PIII cores and thus cheap Xbox contracts.
|
# ? Feb 17, 2017 21:18 |
|
WhyteRyce posted:I thought Intel's deal let them charge the same price throughout the life of the console. So instead of the cost going down over the life of the console, Intel kept getting to charge the same price for a laughably out-dated CPU That's very normal for console CPUs. The initial price usually is less than what the CPU maker could have had, but then they get to keep the same price for 5-10 years of manufacture and come out ahead once production really ramps up.
|
# ? Feb 17, 2017 21:24 |
|
wargames posted:is that good or bad? My i7-3820 @ 4.0 GHz got 1422 and 6458 for single and multi threaded respectively.
|
# ? Feb 17, 2017 22:09 |
|
I7 7700K @ 4700 w/DDR4 3000 2337 and 9925 Looks like Ryzen has the upper hand with multithread and loses single thread. Based on clock speeds I am very impressed with heat/vs core count and not impressed by single thread performance at all. redeyes fucked around with this message at 01:34 on Feb 18, 2017 |
# ? Feb 18, 2017 01:26 |
|
A 3.5GHz Ryzen vs a 4.7GHz 7700k isn't the best comparison though, we will have to wait to figure out how high Ryzen will clock.
|
# ? Feb 18, 2017 01:59 |
|
Hmmmm, my 5820K@4GHz w/ DDR4-2400 CL12 does 1781 and 11292.
|
# ? Feb 18, 2017 02:30 |
|
MaxxBot posted:That's better than what I'm seeing from a 4.5GHz 6700k so I'd say really, really good. I have no idea what that benchmark consists of though. Not in single-thread. I got 2000/8778 on my 6700K at stock (their reference values say 2031/8554). That said, it's clocked a bit higher so we'll see how close single-thread scales clock for clock.
|
# ? Feb 18, 2017 02:43 |
|
Just hoping it can hit at least 4.5GHz reliably. Much less then that - meh. I really do wonder about how fast the memory controller will clock also. 6600K @ 4.4Ghz / 3600 Ram scores 2234 / 8660. GRINDCORE MEGGIDO fucked around with this message at 03:07 on Feb 18, 2017 |
# ? Feb 18, 2017 03:03 |
|
Stock 5775c is getting 1772/7500~ 4.1 OC gets 1971/8484~
|
# ? Feb 18, 2017 03:05 |
|
If and only if it scales perfectly clock for clock, which is highly unlikely, then at 4.2 GHz (stock 6700K turbo) it'll bench 2265 in single thread. Following the same idealistic points/MHz scale, at 4.8 GHz it would score 2589 in single thread.
|
# ? Feb 18, 2017 03:18 |
|
Ye old 3930K at 4.6Ghz with 2133 DDR3 gets 1734/11046 which puts it on par with a 5930K at least.
EdEddnEddy fucked around with this message at 02:15 on Feb 21, 2017 |
# ? Feb 18, 2017 03:33 |
|
My L7500 @ 1.6 GHz does 214/425.
|
# ? Feb 18, 2017 03:40 |
|
This appears to be a very large upgrade over my Pentium MMX
|
# ? Feb 18, 2017 04:09 |
|
Phenom II X6 @ 3.9ghz is 771/4211 Guess I need to upgrade sometime
|
# ? Feb 18, 2017 04:50 |
|
My overclocked i5 750 is good for 1249/4355
|
# ? Feb 18, 2017 05:49 |
|
my abacus couldn't even compile the benchmark
|
# ? Feb 18, 2017 05:59 |
|
1208/7525 out of the FX-8350 lol
|
# ? Feb 18, 2017 06:10 |
|
VostokProgram posted:my abacus couldn't even compile the benchmark I'm just imagining a whole room full of tiny Chinese men worrying away on abacuses for weeks, hand calculating each instruction. A brilliant young one comes up with a mechanical apparatus that allows them to do mechanical AND, OR, NOT and NOR via an incredibly complex series of levers. He is beaten to death by the others because they don't want their cushy 2 year contract to disappear.
|
# ? Feb 18, 2017 06:58 |
|
Wanna see something inflammatory? https://twitter.com/VideoCardz/status/832583990284517381
|
# ? Feb 18, 2017 07:37 |
|
That is pretty funny. Also that first gen i5 beating "modern" AMD's lol. Thank God Ryzen appears to be a break from that trend. Do you think it could be possible they are seeing it OC to 5Ghz on Air/Water? Or OMG this thing doesn't OC at all? (I wouldn't imagine they would have wide eyes and be "Promoting" this if it was garbage.
|
# ? Feb 18, 2017 07:41 |
|
|
# ? Jun 10, 2024 11:01 |
|
The clarification is they're actually impressed with the score at stock.
|
# ? Feb 18, 2017 08:05 |