Quantcast
[ 3 / biz / cgl / ck / diy / fa / g / ic / jp / lit / sci / tg / vr / vt ] [ index / top / reports / report a bug ] [ 4plebs / archived.moe / rbt ]

Due to resource constraints, /g/ and /tg/ will no longer be archived or available. Other archivers continue to archive these boards.Become a Patron!

/g/ - Technology


View post   

[ Toggle deleted replies ]
File: 23 KB, 700x409, 1bfc6039-3afd-4e62-b96f-74fb8f9a5a01.png [View same] [iqdb] [saucenao] [google] [report]
59057780 No.59057780 [Reply] [Original] [archived.moe] [rbt]

What are your thoughts about the Ryzen Sr3 1200X? It's gonna be the i5 killer... or the i3 7350k counterpart

>> No.59057795

No, it's gonna be the i5 executioner if the newest leaks are correct.

1337 stock cooler + unlocked multiplier from the get go.

>> No.59057817

>>59057795

Dude, shut the fuck up with your faggotry.

You're overhyping.

>> No.59057863
File: 47 KB, 640x779, AMD-Ryzen_R3-R5-R7.jpg [View same] [iqdb] [saucenao] [google] [report]
59057863

>>59057795
The r5 1400X its the equal to i5 7400, i'm just afraid to know where is the r3 or if is gonna be shit like the athlon x4 860k

>> No.59059171

>>59057817
The hype train cometh

>> No.59060047

>>59057863
thats price wise nigga

>> No.59060079

>>59059171

They're going to build up too much speed and derail the mother fucker like they did with Polaris.

>> No.59060197

I don't see the point in paying for any of the "X" CPUs for the auto overclocking. Why not just increase the turbo and end up with a better result? Basically the X CPUs are for tards unless I'm missing something

>> No.59060213

>>59060197
>I don't see the point in paying for any of the "X" CPUs for the auto overclocking. Why not just increase the turbo and end up with a better result? Basically the X CPUs are for tards unless I'm missing something
You've got it right. The X cpus are basically for dumb people who want e-penis points.
>muh binning
speculation

>> No.59060224

>>59057863
>R3 1100
>OC's to 4 GHz with stock cooler
>CPU+mobo combo is $175 at Microcenter
my body is ready

>> No.59060473

>>59057780
Shit, I intend on getting a 1700x to replace my Xeon.

>> No.59061480

>>59060213
The X versions don't come with a cooler and i believe it's made with the best silicon. All ryzen cpus are unlocked, why not take advantage of that.

>> No.59061497

>>59060224
Me too dude, the only thing that suck is the the pair of DDR4 prize are more than 120€ for 16GB

>> No.59061499

>>59057817
And here we have the insecure scared Intel sheep.

>> No.59061513

>>59060213
They are binned moron look at the tdp

>> No.59061522

>>59061513
>They are binned
All CPUs are binned, ledditor. TDP could be higher because it overclocks automatically over 4GHz.

>> No.59061525

>>59060213
>>59060197
XFR is not auto overclocking.
It gives you free frequency at any power level, no matter what your base clocks and turbo pstates are set too.

>> No.59061547

>>59061525
What's the advantage of having XFR over overclocking to 4.5+GHz?

>> No.59061551

>>59060473
Do you have a Nehalen Xeon?

>> No.59061556
File: 45 KB, 669x760, AABslX.png [View same] [iqdb] [saucenao] [google] [report]
59061556

>>59061547
You overclock your Ryzen chip to 4.5ghz.
XFR will push frequency even higher than 4.5ghz whenever possible.

The chip knows the characteristics of the transistors its made from. It knows how they behave at a given temp, level of leakage, etc. It is monitoring these metrics in real time. The chip internally adjusts settings to squeeze out a tiny bit more frequency so long as it has the thermal headroom necessary to exploit these characteristics. Its free frequency at any level.

>> No.59061566
File: 28 KB, 407x400, CPU-Z.png [View same] [iqdb] [saucenao] [google] [report]
59061566

>>59061551
SB-E Xeon, although I do have a pair of 32nm Quad-core LGA 1366 Xeons (1 in use)

>> No.59061641

>>59061566
Good change, but what about the quad-channel in the ryzen, they're gonna put that feature right?
Maybe in the best mother boards?

>> No.59061656

>>59061641
Ryzen will be dual channel, but the bandwidth change from quad channel DDR3-1600 to dual channel DDR4-3200 should be negligible.

>> No.59061674

>>59057780
I will be honest.
I don't think the Ryzen 3 1200X or Ryzen 5 1400X will even beat the i3 or i5 from Intel.
The only CPUs from Ryzen that are worth to buy and look into is the Ryzen 5 1600X, Ryzen 7 1700X

>> No.59061711

>>59061656
That would be an interesting one, since the peak theoretical bandwidth of each is an identical 51.2GB/s. Real world performance always comes down to the memory controller itself though. Still somewhat of an unknown quantity for Zen.

>>59061674
>I don't think 4 physical cores with similar IPC and clocks will perform better than 2 physical cores +2 virtual threads that only provide a 30% uplift in multicore scaling
You're entitled to your opinion. No matter how dumb it is.

>> No.59061725

>>59061711
Don't be butthurt now, until benchmarks come out for the rest CPUs. I will see the CPUs as not good at all.

>> No.59061767

>>59061725
Are you passing your native language through google translate or are you fucking retarded?

>> No.59061776

>>59061767
>replying to poojeet

>> No.59061838

>>59061556
There's no auto-overclocking system in existence that can detect and correct instability in real time with zero performance hit. If XFR cranks frequency beyond what you find is your max stable overclock, it'll just cause instability. There's no free frequency to be had beyond the max that your chip can handle, and if you're any sort of overclocker, you'll be pushing fairly close to that limit anyway.

I'd imagine it'll either be automatically disabled or highly recommended to be disabled when you overclock manually. I remember having to disable Turbo Core when I cranked my 1090T up to 4GHz, as it'd take things even further and the chip couldn't handle it, even on a single core.

>> No.59061851

>>59061838
Those are awfully nice fallacious assumptions you're making.

>> No.59061860

>>59061851
That's an awfully nice way of saying "I'm tech-illiterate and can't refute anything you say, but I strongly disagree."

Do feel free to actually argue any point I raised. Be sure to provide actual facts in the process.

>> No.59061867

>>59061838
I think the real advantage will be in 'unexpected' workloads when there's still plenty of thermals left to crank beyond what a normal overclock would be at. Normally your overclock is set to what you can operate at 100% for a extended period of time. To accomplish this you can't just set to +500Mhz, because the thermals overhead may not be there. But XFR can boost past your preset for 100% for the amount of time it takes to warm the heatsink before the effects of diminishing thermals come into play.

>> No.59061912

>>59061860
You didn't make a point, you asserted something as factual when you know literally nothing about the topic. You're talking out of your ass, and now you're instantly defensive because you know you don't have a leg to stand on. Its what all shitposters do.
XFR is not auto overclocking which is something you don't understand. It adjusts metrics that the motherboard and BIOS do not expose to the end user. It is exploiting material characteristics of the transistors themselves within a certain temperature range to squeeze out small frequency gains. In fact the end user has no control over it whatsoever, and the chip does in fact check if its target XFR frequency will be stable, in real time.

>> No.59062270

XFR, if delivered as advertised, will be a highly dynamic resource scheduler. The resources in question being voltage and clockspeed potential.

AMD claims it can sense when a workload is biased towards certain cores and clock them higher independent of others on the fly to the maximum your cooling solution and the chip's voltage handling ability can provide. Thus the chip doesn't sacrifice single core speed for multicore or vice versa.

Considering cpus and gpus can already downclock and reduce voltage when idling, I imagine XFR will expand upon that as well, coming as close to "switching off" unutilized cores as possible while delivering maximum resources to the cores being taxed hardest.

This is also done on a HARDWARE level. There's a considerable portion of diespace dedicated to this feature, and for good reason.

Beyond being a major selling point, it allows AMD to easily and quickly test each chip to know its actual potential. Thus far, it would be necessary to test each chip in a number of configurations, determine how viable these targets are, then set product tiers based on limited data points and leave the rest to chance. This is the basis of the "silicon lottery." It's why we sometimes get exceptional overclocks and sometimes get total turds. It's in the manufacture's best intrest to account for all chips and price accordingly, but extensive testing adds significantly to cost.

What I see XFR as is a simple method to automate the binning process. AMD can instead gather datapoints from all chips produced and determine which ones are best, which ones are worse, and everything between.
That said, it doesn't mean non x and lower tier chips cant benefit from manual overclocks. AMD could have a target like "x clockspeed at x frequency for x amount of chips" and lower binned offerings could be lacking in just one area. We don't know those targets, however, so the lottery continues.

>> No.59062277
File: 232 KB, 1097x186, ryzen.png [View same] [iqdb] [saucenao] [google] [report]
59062277

First NDA for prices expired

>> No.59062305

>>59062270
My question about XFR though is if the end-user can alter the TDP target for the chip and have XFR work within the new power range.

>> No.59062316

>>59062270
Fuck meant to say "x clockspeed at x voltage on x ammount of cores."

I fucked that all up

>> No.59062358

>>59062305
Tha we don't know yet, but I'm sure it will be at least touched on in reviews next week.

If we're lucky we'l get options in oversride to set thresholds and mabe even quick toggles for things like "cool and quiet mode" and "high performance mode."

If we're not lucky the specifications are also on a hardware level and cannot be bypassed without manual overclocking.

>> No.59062409

>>59061674
>I don't think the Ryzen 3 1200X or Ryzen 5 1400X will even beat the i3
Um... the 1200X has twice the cores of an i3-, and the 1400X has 2-4x the threads.

And the i3s are mostly clocked pretty low except for the i3-7350k which is a 4.2Ghz dual core that costs $30 more than the dual core 1200X.

Hyperthreading is simply not better than twice the cores unless the per core performance is significantly higher.
But since the 1200X will probably overclock to well over 4Ghz, and the IPC is within 5-10%, the per core performance of the i3-7350k is not "significantly higher".

The only point of buying intel at the moment is for the G4560. But there are some games that have been coming out that pretty much don't support dual core anymore.

>>59061867
Also overclock also includes your idle wattage and temps.

It's yet to be seen how well XFR works, though.

>> No.59062450

>>59062277
In the video it says the XFR will only bring the 1800X that is 3.6ghz-4.0ghz to 4.1ghz. That's fucking nothing.

>> No.59062466

>>59062450
As I get XFR plays within the allowed TDP range

>> No.59062485
File: 226 KB, 1920x1080, recordbeforerelease.jpg [View same] [iqdb] [saucenao] [google] [report]
59062485

Reminder that AMD beat the OC record

>> No.59062548

Will there be some kind of Pentium G4560 equivalent or are they fucking over the poors?

>> No.59062556

>>59062548
They'll have cheaper APUs.

>> No.59062574

>>59062548
The lower tier stuff is coming later, like how new processor releases are supposed to go.

Release the best stuff first to get the most profit out of those wanting to get the new tech, then release lower level versions, cashing on the success of the higher ones.

Releasing lower tier stuff first is a sign that the higher-tier stuff is garbage.

>> No.59062584

>>59062548
>>59062556
There will be 4c/4t Ryzen CPU's. (Not sure if they are APU's as well)

>> No.59062692

>>59062466
If it played within TDP range then it should go to 4.5Ghz-5.0Ghz with a custom loop, not only 4.1Ghz.

>> No.59062719

I didn't follow anything lately, when is Ryzen going out to markets?

Also, i guess comparisons are in favor to Ryzen, right?

>> No.59062734

>>59062305
If we can do that in software like we do for GPU's I'll cum all over my face and sell my mother's gravesite to get Ryzen

>> No.59062754

>>59057780
i5 killer

>> No.59062756

>>59062692
I mean the 95W limit specified by AMD. a better cooling like a custom loop would increase the TDP headroom considerably but like you say it doesn't seem to be auto boosting to take advantage of that, pretty useless if it's not configurable

>> No.59063018

>>59057863
Correct me if I'm wrong, but isn't 1400x 3.5-3.9+XFR? that would put it next to i5 7500 in my opinion.

>>
Name (leave empty)
Comment (leave empty)
Name
E-mail
Subject
Comment
Password [?]Password used for file deletion.
Captcha
Action