Explain me why is or isn't multithreading important bec

>post tagged under RUMOR

>>58704306
if they sell it for i7 price, soon everyone going to own 1.

>>58704306
U r reatarded lamo

>unsourced anonymous rumors machine translated from moon runes
Nope. The upcoming lineup is
8c/16t
6c/12t
6c/6c
all with two different clock variants priced from $200-250 to ~$600. Only 4 core Zen CPUs released will be the rejects from Raven Ridge APU production coming out later this year, the Summit Ridge is their HEDP only lineup.

>>58704306
>4c/8t at 45w tdp
it's over, intel is finished.

>These chips will feature a base cock of 3.4GHz or higher

Have we gotten any non-biased real-world benchmarks of the Ryzen CPU's yet?

>>58704905
yes

>>58704905
yeah same as when Pentium 4 came out.

>>58704978

>translation: they were unbeatable

>>58704905
CPC Hardware did some on an ES, and I wouldn't call them particularly biased.
Their benchmarks put Ryzen clock for clock between Broadwell and Skylake.
Keep in mind this was an early engineering sample with SMT and uop cache bugs.

>>58704848
No? AMDs 65W quad cores (x4 845) are total crap. Why should their new 45w quad cores be that great? Probably they will be in competition with dual core i3s, since they will be marked as SR3 and not under the SR5 or SR7 name.

>>58705110
ipc increase, it's not about muh cores.

>>58704306
>new Ryzen cpu won't support them.
>8c/16t
>wont support
>8c/16t
U dun goofed

>>58704905
No because no one really has a chip yet

>>58704306
you know a rumor is bad when the rumor directly contradicts what amd has said.

base 8 core 16 thread is at most 500$, as amd has said at ces, its either going to match i7 price, or its going to go for the 6 core price.

amd not able to turn off single cpu segments or at least 2 segments at once, needing to ditch an ENTIRE half of the cpu if its fucked, that screams horrible design, something that shipwrecker is not known for.

smt disabled offers a range of better costing cpus, from 8/16 8/8 6/12 6/6 4/8 and 4/4 where they have a FAR better way to tune for price with 2 or 3 binned chips each segment that have different guaranteed clocks. this offers you a range that goes from at worst, 500$ for lowest 8/16 down to 100 for a completely fucked die 4/4

But I honestly believe amd will match ball park i7 price for 8/16 and have a higher binned chip for the 500 or 500+ market. maybe even just segmenting the consumer market with a crap die 8/16 and a good die 8/16 and put out workstation mcm 2 die 4/8 cpus so they have no disadvantage other then avx 512 and charging 700-900$ for it.

>>58705110
amd is about equal if not better ipc then broadwell, skylake was not a massive gain on that, and kabby is literally the same.

2 cores 4 threads get their shit kicked in by 4 cores, not to mention 4 core 8 thread.

onyl reason the i3 is seen as an option is some are overclockable now, and when oc'ed they can surpass an i5, all amd chips are oc ready though, and with no upper limit to the turbo, possibly a fairly high out of box max clock too if you get a good cooler.

Honestly, if amd sells the cpus on their base clock, clamp down on only good motherboards can overclock, and then have something that can figure out the max clock, amd is going to look fucking amazing compared to intel in the mass market.

>>58704836
everything besides the 8/16 16/32 and 32/64 is a rumor, as amd has yet to say anything about any of these besides ryzen, the other two were only found out due to leaks from partners and samples given to cern.

as for no quad core, its possible, but its also possible there are places on the die, if there was a defect it would effectively kill one cpu cluster. so its entirely possible we will se quad cores, though honestly if thats not an issue, it would be hard to imagine a cpu that is so fucked half of its broken...

Its also possible horrible binned chips could just have 4 core disabled or cut, just so it can fit in a power range,

>>58704905
>>58705379
>>58705024
Yes and no.
There is one frech place that got a 3.15 es and benched it, apparently it wouldn't run over 3.2 with the turbo, so it was literally a worst case scenario for the cpu lineup.

Shit is VERY promising, as just the .25ghz higher clock alone puts it between i5 and i7 for games, and within 10% of intels 8 core, all while throwing a fuck ton of errors

assuming everything works and there are no errors in the final product, price will determine how we receive this cpu.

>>58704306
An 8c/8t part has roughly 12% more compute power than a 6c/12t, is probably as efficient or more, likely has better multithread scaling, and is likely just as cheap to die harvest. We have to take into account a 6 core zen cpu will probably, very probably have contention and data transfer issues between the CCXs compared to a full 8 core part. That theoretical 12ish percent is likely more in real practice.

to bring you guys back on topic.

Ryzen will NOT support multithreading, and back to OP's question: Who needs multithreading

>>58705745
>Ryzen will NOT support multithreading,
How do you explain the 8 cores?
How do you explain the 16 threads?
Everyone needs multithreading.

>>58705745
What do you base this claim on? You realize AMD has stated in front of investors that their chips will support SMT right? You realize that if they lied they will be sued by their investors and their credibility will be completely ruined right? This doesn't just apply to AMD the company it also applies to the executives who are running AMD including Lisa Su.

I think it's easier to assume Ryzen will have parts with SMT than to assume AMD is just going to fuck themselves up.

>>58705892

rumors are from varoius sites and why should they spread bad rumors about the new cpu not supporting multithreading?

>>58706028
For ad review? This is part of the reason why /g/ hates currytech, they're a damn rumor mill

can anyone give me a good example how multithreading works?

For example clicking a link in skype thats opens my browser. Is that it?

File: KHAAAAAAAAN!.png (690KB, 726x799px) Image search: [Google]

690KB, 726x799px

>>58704306
>no 6 core parts

>>58706798
No

>>58706817
you get 8 core version for 6 core price

>>58705560
>There is one frech place that got a 3.15 es and benched it, apparently it wouldn't run over 3.2 with the turbo,
Is this the same French site that got one core running at 5 GHz?

Sidenote, why hasn't AMD gotten turbo to work yet? Every AMD chip I've had so far performs better with it off.

>>58707154
No evidence of that.

>>58707532
its AMD this should be evidence enough

>>58707266
turbo on this will likely work, but I see turbo in general as fucking stupid.

as for if its the same place, dont know, as after zen comes out they will be irrelavant again.

>>58706028
you clicked the site, even though you are 1/3000th a penny to them per ad they still make a few dollars off the web page loading just that one time.

>>58706798
look at a gpu, their cores are basically retarded but for the task they need to do they get the job done, so all those cores work in tandem to make the pretty picture.

the more complex something is the harder to multi thread though, its why you see more things with good gpu acceleration before they ever have good multithreading.

>>58709407
>its why you see more things with good gpu acceleration before they ever have good multithreading.
What the fuck am I reading

>>58707532
ces, only 2 price points for base 8 core are price matching i7 or 500$, those are the only two they were considering, and honestly I think they will go i7 because they know damn well how big that would be to computers, and how many sales that will generate while intel has no answer, and if they don't modify cannon lakes price and down shift the i7e platform, will continue to have no answer in a years time.

>>58709435
programs that are easy to multithread are also very gpu accelerateable. where as shit that's hard generally does not. look at chrome, the way this shit made multithreading was by instancing itself, a VERY basic program that isn't easy to thread but would greatly benefit from it.

>>58706817
the rumor is retarded, the FX 63xx chips were probably the most popular in the line up

>>58704306
>Explain me why is or isn't multithreading important because the new Ryzen cpu won't support them.
You write like a fucking retard. God you are stupid, just kys!

>>58704836
Fucking 200?might as well go i5 7500 for 220 bucks