How is ryzen suddenly faster in Tomb Raider on Ryzen? What

File: Screenshot from 10-04-2017 23:42:55.png (124KB, 1201x831px) Image search: [Google]

124KB, 1201x831px

Apparently the Taichi is too perfect, as /g/ has been saying since launch.


ASrock confirmed bestest

>>59831386
>>59831351
>Taichi loses in all but one benchmark
>DAH BEST
exepct no less cherrypicking from AMDumbfucks

>>59831351
There's varying difference between motherboard EFI/BIOS revisions, especially on AM4

>>59831495
Fantastic reading comprehension you monkey.

>>59831495

You sound anxious and aggravated, how come?

Do you have a pathological need to have someone view the world around them the same as you?

>>59831351
TPU are pretty huge Nvidia shills though.

>>59831495
>he literally doesn't understand the difference between a motherboard review and a CPU review

Intel customers, everybody.

>>59831507
>>59831507
>>59831507
This should end the thread.

>>59831495
you don't even bother to click it. do it?

This RGB LED bullshit has to stop.

>>59831637
The third paragraph of the conclusion mentions minimal LED nonsense

>>59831386
Based ASrock.

>>59831386
He forgot to mention the Taichi actually supports ECC memory.
As in error correction actually works.

>nowhere to be found

welp

it's so good it's selling like hotcakes

File: Screenshot from 11-04-2017 00:07:38.png (125KB, 1017x300px) Image search: [Google]

125KB, 1017x300px

>>59831351
>>59831386
This is the 1800X review from the same site from a month ago.

Pretty hilarious.

>>59831386
>only one M.2 slot
>perfect.

My Assrock has 2 M.2 slots
So it is more perfect, it is perfecter.

>>59831757
There's two, but one is 16Gb/s

Might wanna actually read next time.

>>59831773
>Might wanna actually read next time.

Great advice, will do.

https://www.techpowerup.com/reviews/?category=&manufacturer=&pp=10000&order=date

There's literally 5 10.0 scores on TPU since 2004.

The last one was in 2014.

>>59831889
What are you trying to imply?

>>59831916
It means they really like it.

>>59831744
>>59831351
BRAND NEW PLATFORM RUNNING NONSTANDARD MONOLITHIC SHARED L3 ARCHITECTURE ISN'T OPTIMIZED AT LAUNCH DAY


STOP THE PRESSES THIS IS SCANDALOUS

>>59831942
Good luck telling that to Intelfags.

>average intelcuck: >>59831495

>>59831351
>What the fuck is going on with this platform?
Pajeets are being used as paid beta testers, that's what.

Which is best ~100$ mobo?

>>59832003
Some 200 man internal validation team can't begin to compare to millions of faggots with the pure amount of data churned

>>59831351
Because you are trusting a website that has UNFAIR reviews, aside from their inability to proper bench stuff.

>>59832012
Asrock AB350 Pro4

There's a mATX version as well, same everything besides size

>>59831351
bugs.

>>59832040
This unless you can find a $100 board with a Intel NIC and 7+ chokes.
If you're not overclocking you can live with 3-4 chokes

File: 1490472047431.jpg (599KB, 1312x1740px) Image search: [Google]

599KB, 1312x1740px

>>59832022
>>59832003
You again?

>>59832040
Can you overclock on it? I mean it only have 3 phases(doubled to 6)

>>59831351
amd finewine already at work

File: AB350%20Pro4(L2).png (437KB, 1200x1000px) Image search: [Google]

437KB, 1200x1000px

>>59832087
It has 8 phases, what are you talking about?

That's 80% of the reason why people recommend it for the B chipset boards.

>>59832111
9 chokes and 8 caps, for a $95 motherboard that's fantastic power delivery.

>>59831351
>this performance with 80% higher DRAM latency than Intel

What will happen once the IMC drops to 50-60ns?

>>59831351
AMD sucks
>get ahold of daddies creddit card
>want to buy a new pc
>hear my reddit buddies talking about Ryezin SeePeeYou
>made by AMD? Ewww.
>they say it's good but the platform is new
>don't really understand so I buy an 1800-ex and a major gamer Giguh-bite board
>Micro 80x board that has no reviews yet becuase I simply can't wait. I also preorder all gaymemes
>I need my computer to be portable, how else will I show others just how underground I am with my Whyzn?
>board arrives and cpu arrives
>Gigabyte board is trash
>problems
>daddy asks to see new pc
>tell him it doesn't work
>he gets visibly upset at me for wasting his money
>tell him it wasn't my fault, AMD is BAD and tricked me
>he doesn't care, no good boy points for the rest of the week
>swear i'll never buy amd again and instead buy a used 5960k for 700 USD
>because I need something reliable and intel never has problems

A-Another massive disappointment, am I right /g/uys?

>>59832226
The MUH GAYMING fags would stop shitposting

>>59832134
Can someone fill me in on what I'm physically looking for on the actual board to know this stuff? Still new to anything beyond sticking a locked CPU in a cheap board and calling it good so it's not something I've ever had to learn.

>>59832111
>>59832134
Isn't it 9 phases? 3 phases for soc and 6(doubled from 3) for cpu?

Other manufacturers offer atleast 4 phases for cpu.

>>59832238
Can't tell if pasta or trolling

>>59832257
But Ryzen is a SoC.
If you meant 3 phases for the PCH/SB, then that's nonsense, it uses some 8 watts.

>>59831546
This made me laugh way more than it should have.

>>59832266
It's pasta I just made from witnessing some jackass.

The end result actually happened

>>59831351
I've been waiting for this motherboard to be in stock since the beginning of april

>>59832295
for future gpu in apu?

if ryzen wasn't powerful, intel wouldn't have SLASHED their prices in response to its release

intel knows a lot more about what the chip is than shitty benchmark sites

>>59831386
all of the amd "high end" mobos have only 8 real phases, including taichi.

>nobody did any interrupt testing on 6900k vs Ryzen


Why is this? Why did nobody load the 16 threads to 100% on both systems and see the responsiveness of the system? On my Intel 4 cores when I peg them at 100% the system is literally unusable, it can barely process interrupts from the fucking MOUSE.

>>59832249

A world of information at their fingertips and babies still need to be spoonfed.

>>59832353
Intel doesn't slash prices, period.
They didn't slash when K8 came, they will never slash now.

For fucks sakes this is the company that admitted in its earnings call that it would rather take a marketshare hit than lower their precious 60% margins.
They're trying real hard to be 'premium' like Apple

>>59832369
>when even my ancient Phenom 2 tricore will handle mouse movement while pegged

Only thing that fucks it over is memory IO because I used DDR2, that's when the system freezes up

>>59832387

I don't think companies understand that Apple has almost gone bankrupt more than once.

>>59832362
How is that not enough for anything but LN2?
You know there are 20A+ chokes, right, pure choke count isn't all that important, but more is almost always better in this case.
Not to say you can't have 4 expensive chokes doing the work of 6 cheaper chokes, because you can.

Who actually got tears in their eyes from seeing AMD back in action? Their last real releases have been Athlon and Hammer.

I'm so fucking happy.

>>59831744
>>59831942
The test in OP is for some reason running at 1920x1200, making direct comparison pretty worthless. Like, surely if you're a benchmark website you'd have some standardised test methodology. Lazy shits.

>>59832824
There's little difference between 1200p and 1080p, and even with Zen doing better at higher resolutions, those 120 vertical lines surely won't make it suddenly 13% faster than Kabylake who it was losing to by 15% a month ago.

>>59832859
You saw those very inconsistent results over the last month?

94% of that is down to motherboard EFI differences, rest is Windows.

>>59832369
You can't possibly expect gaymerfags to know about such intricacies.

By the way, I have no such problem in my poorfag APU build, and all construction cores handle interrupts really well. Don't know why it would be different with Ryzen which, despite being an all-new architecture, is an evolutionary step over Bulldozer.

I've been out of the loop. Are we saying the Taichi is the best AM4 motherboard or just X370?

And what about the other big boy ones like the Crosshair Hero?

>>59833008
Taichi is currently the best AM4 motherboard, period.
Crosshair Hero on top of being expensive as shit, had a hilarious amount of bugs on launch, I don't know how it's performing now.

>>59833008
Asus really fucked up with Ryzen. All of their boards have some teething issue, from instability to shipping outdated BIOSes.

>>59833031
>>59833058
And for the ~$150 range? What's good there? I have an Asus PRIME X370 sitting in my cart and while I know they kinda flopped with the Ryzen stuff like you said, it doesn't seem like anything that can't be fixed? Just a case of making a bad first impression.

I'm considering holding off until a Taichi is in stock though.

>>59833126
To be honest the only two choices I'd consider is the $200 Taichi or the $100 AB350 Pro4

Nothing in between caught my eye.
Losing the X chipset means absolutely nothing to me as I don't care about SLI/xfire

>>59833151
So where do I buy a Taichi?

File: Screenshot from 11-04-2017 02:11:36.png (378KB, 565x939px) Image search: [Google]

378KB, 565x939px

>>59831351
>>59831351
Current benchmarks are fucking USELESS after the AGESA update.

Look at this, a latency reduction on everything across the board.

>>59833407
I can find some where I live, but that's not in the US.

File: 1490031269751.jpg (19KB, 140x140px) Image search: [Google]

19KB, 140x140px

>>59833408
>26% faster mem access
>18% faster L1 access
>17% faster L2 and L3 access

>from a software update

THE CALL THAT SAVED AMD

>>59833581
It's a microcode update, which is technically firmware. Which is technically software but not really.

>>59832003
This is what a new arch launch looks like every fucking time you deluded cocksucker.

>>59833721
Yeah, everyone tests a new architecture on early adopters, it's how it works.

Intel's x58-x99 launches haven't been exactly stellar either, and they're mostly similar CPU designs with some tweaks

File: LELELE.jpg (66KB, 630x218px) Image search: [Google]

66KB, 630x218px

Thank god I waited.
Enjoy your 1 year lifespan ryzen firehazard stations, SKL-X out in 3 months

File: Screenshot from 05-04-2017 23:59:51.png (94KB, 879x714px) Image search: [Google]

94KB, 879x714px

>>59833909
I like how Sky-X has literally no actual purpose.
Not even the PCIe lanes, nor 4 channel memory controller now.

>>59833909
nice meme retard, Ryzens are the most power efficient cpus up to date

>>59833909
>entire platform existence invalidated by a EFI update and a MCM

Lmao

>>59833408
wouldn't this fix this the slow ram performance loss?

File: Screenshot from 02-04-2017 21:00:52.png (39KB, 826x775px) Image search: [Google]

39KB, 826x775px

>>59834233
Alleviate yes, but not 'fix'
There's really nothing to fix, testing showed that there's not all that much difference between 2+2 CCX vs 4+0 CCX configs, some 10% at max

Still, even Ryzen benefits from higher clocked memory, so that's a cheap upgrade path down the line if you have cheap memory now.

File: aida_latency.png (16KB, 500x250px) Image search: [Google]

16KB, 500x250px

>>59831351
Just a reminder that this was ran without the AGESA update.

>>59833408

>>59834251
than what's the cause that ryzen demands fast ram to work proper?

>>59832249
In the same time you spent with your frustrated typing, you could have simply just shared the knowledge, clearly you know nothing newfag.

>>59834316
Nothing? It's a bandwidth hungry architecture.
Not to say some parts like the fabric don't benefit from higher memory speeds which in turn benefit gaming performance, but generally the CPU itself wants bandwidth, probably a design decision and not a bad one considering DDR5 is out in the next 2 years.

>>59834339
>nothing
>it's a design decision

i will pick number 2. thx. i'm excited for the upcoming benchmarks

>>59834424
And as a design decision, it has its drawbacks, which being slightly slower performance in some applications. But as a design decision it also has its pros, which is efficiency and scaling, all thanks to that 'problematic' infinity fabric.

Can we talk about the massive GENOCIDE that AMD will bring to Intel with their Naples?
Seriously lads, you think that AMD could gain some server marketshare again?

>>59834467

No, because Intel is already trying to lock data centers in with sweet heart deals.

>>59833909
>1% more performance for the double of the price a bang for the buck
>130 Watts, goy, you save in heating cost, hehehe

>>59834503
Could this be bring to the US DoJ?

>>59832012
Gigabyte because of BCLK chip

File: Screenshot from 07-02-2017 21:15:31.png (447KB, 1877x957px) Image search: [Google]

447KB, 1877x957px

Intel Corp. still has the most advanced foundries on the planet.
Intel Corp. is dedicated to creating high performance products using non-conflict materials, which we are the world leader in.

>>59835142
Intel Corp. "most advanced" foundries on the planet produce a slower 8 core CPU that uses more power at idle than its competitor 8 core uses at load.

Rather say you have shit fabs instead of a shit architecture, it's less embarrassing since for fabs you can throw more money at the problem, for architecture you actually need skilled and smart people.

File: 1490348833037.png (30KB, 454x534px) Image search: [Google]

30KB, 454x534px

>>59835207
Reminder that the 1700 is closer to the 6900k's IDLE power consumption than its load one.

>>59831351
most motherboard manufactures are getting lazy and Ryzen is a brand new platform, something we havent had in almost a decade, especially from AMD

>>59831386
>10 rating for a motherboard

When is the last time TPU gave a 10? I literally cannot remember, some 5 years ago?

>>59831351
any B350 pro4 mitx reviews yet?

>>59835439
I barely saw any B chipset reviews at all, everyone's salivating over the overpriced ASUS Hero garbage or Taichi(which is actually worth salivating)

>>59833126
I run an Asus Prime x370-pro. It did/does have some teething issues, but nothing that renders the board unusable. BIOS updates come frequently, mostly solving issues but occasionally complicating things. (Higher overclocks were easier in earlier versions, now low-power medium OCs are easier, go figure.)

Currently running my 1700 at 3.6ghz/1.166v with shitty 2133hz ram overclocked to 2400hz (2666 is almost stable, hopefully it will be next bios update.)

>>59835521
If you buy something like 3000 or 3200 RAM does it come at a standard speed with the capability of being overclocked to 3000/3200? Or do you have to underclock it while overclocking the rest of your shit?

>>59835521
ASUS dropped the bomb, it would be fine but most reviewers got the Corshair Hero motherboard for review and that shit was buggier than a anthill, some ASUS boards didn't even get updates and are still in launch day hell.

>>59835581
Technically all DRAM over 2133MHz is overclocked according to JEDEC.

But you can OC pretty much everything, so a 2133MHz one will have no trouble with 2400 or even 2600 if lucky.
A 2900 one will have no trouble with 3200.

Usually it's rated speeds +1 tier as a safe OC on 1.35V(I wouldn't go over 1.4V)

>>59831744
>a month ago
well you're a dumb fuck aren't you then

>>59831351
Back to /v/ with you gamerfag.

>>59835713
It's pretty normal for performance to increase that much in a month? The AGESA patch hasn't even landed yet on the tested Taichi
>>59834313
>>59833408

>>59835738
This >>59834313 looks like some slowmo 2133mhz bench, but impossible to tell without proper notation
This >>59833408 is what ryzen looks like with updates and higher clocks

You kinda answered your own question?

>>59835849
>This >>59834313(You) looks like some slowmo 2133mhz bench, but impossible to tell without proper notation

Check the OP
2x 8 GB DDR4 3200 MHz
G.Skill FlareX F4-3200C14D-16GFX

>>59835925
>https://www.techpowerup.com/reviews/ASRock/X370_Taichi/7.html
Where's the timings listed? That BCLK methodology seems retarded to me

>>59835980
14-14-14-34

It's on Gskills site, TPU doesn't overclock memory unless it's a explicit memory review

File: Screenshot from 11-04-2017 05:59:02.png (148KB, 1750x738px) Image search: [Google]

148KB, 1750x738px

>>59835925
>>59835980
>>59835993
>>59831351
Literally fuck TPU, they couldn't even bother to update the BIOS

>>59836006
Jesus fucking christ they're running at 1.40 BIOS, literally 3 versions apart, these tests could have been ran with the AGESA updates and faster memory.

>>59835993
https://www.youtube.com/watch?v=G80uG3acg44
So the ASUS crosshair hero with it's host of shit reviews has better latency?

>>59833909
That delidlake is DOA.

>>59836052
>Apr9
Clearly the Hero has gotten over its issues since launch.

But does TPU have any reason to lie about DRAM latency when it gave the damn board a 10.0 review?

>>59831495
>shilling for 7700K for weeks
>angry that ryzen was already better bang for the buck
>now people are fixing their gaymes and it isn't even 2-10% less FPS anymore, it's BETTER at almost every game now
>still WAY FUCKING BETTER AT EVERYTHING MULTITHREADED

go shove your pentium up your ass, faggot

>>59835581
>>59835611
Anon basically said all there is to say, but it's worth mentioning that you should go and check compatible memory (with Asus boards you can access this documentation via the support page) just to give yourself a headache-free experience.

>>59836090
recent reviews up until now still reporting like 20-25% doa
will probably still get the taichi myself

>>59833592
Are you retarded?

>>59835278
>30%
>closer

Thats a pretty huge margin dude.

>>59836581
Do you lack reading comprehension

>Reminder that the 1700 is closer to the 6900k's IDLE power consumption than its load one.

6900k load power consumption = 160.5W
6900k idle power consumption = 70.2W

160.5W - 70.2W = 90.3W
90.3W / 2 = 45.15W
70.2W + 45.15W = 115.35W

That means that if the 1700's load power consumption is less than 115.35W it is CLOSER to the idle power consumption of the 6900k than it is to the load power consumption of the 6900k.

That specific finding records it at 108.9W which is LESS THAN 115.35W. Therefore it is CLOSER to the idle than the load power consumption.

Go eat some dicks.

>morons talking about hardware they shouldn't waste their money on
There's no good games you need this shit for.