[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / r / s / t / u / v / vg / vr / w / wg] [i / ic] [r9k / s4s / vip / qa] [cm / hm / lgbt / y] [3 / aco / adv / an / asp / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / qst / sci / soc / sp / tg / toy / trv / tv / vp / wsg / wsr / x] [Settings] [Search] [Home]
Board
Settings Home
/g/ - Technology



Thread archived.
You cannot reply anymore.



File: AMD.jpg (151 KB, 781x767)
151 KB
151 KB JPG
Many benchmarks seem to indicate that the the new Ryzen 2700x has higher IPC than many of Intels offerings.

This this benchmark mean Intel is Finished?
>>
NOOOOOOOOOOOOOO
>>
>>65554140
>2700X
>3.7GHz/4.3GHz
>only 15.6 more in single core performance against an old gen at 3.3GHz/4.30
>>
>>65554198
>>only 15.6 more in single core performance against an old gen at 3.3GHz/4.30
They turbo to the same clock retard
>>
>>65554224
Wow, it's almost as if I already knew that when I put the turbo values in my post you replied.
>>
>>65554128
Big if true. My 1700 @ 3.8 gets 430 ST and 4559 MT.
>>
>>65554243
Glad to know you are acknowledging your stupidity
>>
File: 7900xark.png (93 KB, 1059x939)
93 KB
93 KB PNG
>>65554198
>>65554224
>>65554243
>>65554304
You guys do realize the 7900x has a 4.5ghz max 1/2 core turbo, right? Though Sky-x has pretty bad single thread performance for intel.
>>
File: 1514940385153.jpg (44 KB, 638x629)
44 KB
44 KB JPG
>>65554128
seems to be only 2.5% away from a i7-8700k at the same clock speed
>>
>>65554128
Don't worry. CPU-Z will get another update to 'fix' this discrepancy.
>>
>>65554606
To be fair, have you ever seen any other tests that showed similar performance for Ryzen? It was a particular piece of code in their benchmark that ryzen just happened to be extremely good at. But nothing actually uses. It was beyond outlier status.

Though there are some very specific benchmarks on sisoft that ryzen seems to be a fucking monster at.
>>
>8c/16t Ryzen nearly matches a 10c/20t Core iNein

Good god, HyperThreading fucking sucks
>>
I wonder how much X470 impacts performance over X370.

Also Ian from AT mentioned that the reason there's no 2800X yet is under embargo.
>>
>>65554993
Got a link to that? I'm going to be so pissed if they drop a 2800x uber bin shortly after launch.
>>
>>65555098
No, I saw it on twitter a few days ago and I don't have a phone to search for it now.
>>
>>65555118
Was it actually from Ian Cutress on his twitter account? I'm all the way back to march and haven't seen anything about it. Or maybe he got told to delete it.
>>
>>65555202
I think it was Ian, it was a response to someone, said something about "2800X's missing under embargo until the 19th" or similar.
If the embargo is so close it's not likely to be a TR SKU
>>
>>65554669
>Be good at something
>OH NO NO NO NO NO we can't be having any of that
>>
2800X is probably awaiting Intels response so they can BTFO of them. They learned their lesson from the 8700K.
>>
>>65555385
probably yeah
>>
>>65555385
If they can pull another 200MHz single core and 200-300MHz all core they'll be very close to the 8 core CFL, probably even better in some cases with tuned memory.
>>
>>65554198
>what is price
>what is tdp
>>
>>65555385
Waiting for it works fine for AMD since when it launches it'll be at its most optimal with fully featured BIOS and microcode improvements and the most memory compatibility.
Secondly they're gonna need to build a decent stock from good dies, that will take a while.

Lastly I think it won't actually cost more than $360, but it won't come with a cooler.
>>
>>65555385
I have my doubts if even a 2800X will ever exists.
>>
>>65555814
>>>65554198
>>what is price
>>what is tdp

These are already public information:
Launch price and TDP

https://www.anandtech.com/show/12642/amd-ryzen-2nd-gen-details-4-skus-reviews-19th
>>
>This $1000 CPU beats out this other $1000 CPU by 1%!
Literally who gives a shit?
>>
File: 1500010756103.png (944 KB, 1228x1502)
944 KB
944 KB PNG
>>65556005
>>This $1000 CPU
Ryzen 2700x will be about $ 330
>>
>>65555864
>2800X exist
Why would it? The difference between the 1700X and 1800X were almost nothing. Just make the best and call it 2700X. Done deal.
>>
>>65554696

SMT doesn't suck. It is very good at workloads and code that is hilariously parallel.

The issue that SMT suffers if there's resource contention (Intel never address this until meshed-topology in Skylake-X)
>>
>>65554669
>>65556139
>mine monero on an AMD processor
>using all threads improves hash/s
>mine monero on an Intel processor
>performance actually goes down unless I use the number of physical cores
Wew.
>>
>>65554128
For reference, 8700k at 4.8GHz only scores under 4.4k multithreaded.

CPU-Z is going to have to update to gimp Ryzen harder again.
>>
>>65556377
If this bench is true than there is only a 2,5% difference with a stock 8700k (508 score) on single core

see link for reference
https://valid.x86.fr/bench/1

Intel should be worried
>>
>>65556434
Why should they be worried? They still got the 640x480 benchmarks which rely completely on memory latency, they'll win there!
Just get a few reviewers to find a board with shit default Ryzen timings and you're good to go.
>>
>>65556434
It looks like the IPC increase is anywhere from 2-9% depending on application. That puts it VERY close to Coffeelake IPC, given that Zen1 was above Haswell and Kabylake and Coffeelake IPC is only like 10% about Haswell themselves.

It seems incredibly likely that Zen2 (not Zen+) is going to be significantly higher IPC than anything Intel has to offer. And likely clock close to 5ghz. And have higher perf/watt.
>>
>>65556515
Had a small convo with a guy about Intel's next uarches, apparently everything after Icelake is a IPC increasing tock, besides Tigerlake, so Ice > Meteor > Adler.

Competition will be fierce, next 4 years will be amazing.
>>
>>65556603
Icelake included as well, it's a tock.
>>
>>65556139
I said HyperThreading, not SMT. HT is Intel's implementation of SMT and it sucks compared to Ryzen's.
>>
>>65554128
Heres the I7 4790k at 4.2 ghz and a 1700x at whatever the fuck its boost is, it really does this time see that intel is actually even more fucked this time around than when ryzen first launched, Single core is similar to kaby lake and coffee lake but at a lower clockspeed
>>
>>65555262
Don't worry pal.
Zen2 is good as Vega just wait was. It totally lived up the hype
>>
>>65556512
It's not our fault you only have the most cheap poolaris model and can't have glorious 8700k oc + 180ti sli
>>
File: 1495358050273.png (495 KB, 1070x601)
495 KB
495 KB PNG
>>65556950
Haswell devils lake i7: 415
Ryzen 1: 370
Ryzen 2: 495
Skylake X: 480

INTEL BTFO
>>
Reminder that Ryzen has some AI branch prediction tuner that helps in benchmarks.
>>
>>65557008
So does Intel (and they've done that even longer). Your point?
>>
>>65555262
Shh into the ovens you go.
>>
>>65557040
Because it's cheating!

t. Intlel shill
>>
>>65555385
2700x official specs are for xfr2 4.35ghz
If there's any merit to that rumor then a 2800x would be a very nice surprise. Pushing it to the mythical 4.5ghz?
>>
File: 13247509457687.png (369 KB, 1437x874)
369 KB
369 KB PNG
>>65557005
>>
File: 1522796911002.png (531 KB, 750x731)
531 KB
531 KB PNG
>>65557385
Wait a second, how is this 8700k working on a Z170???? HOW IN THE FUCK!!!!
>>
>>65557421
You need an unofficial bios
>>
File: 1502055755975.png (343 KB, 672x794)
343 KB
343 KB PNG
>>65557385
Thats cpu z not cinebench favelamonkey
>>
File: Delusional.jpg (14 KB, 938x120)
14 KB
14 KB JPG
>>65557464
>implying anyone but a diehard intel fanboy is going to go through any of this bullshit
Keep fucking dreaming
>>
>>65556005
>2700X
>$1000?
state of shills, jessus christ..
>>
>>65557533
He must've read that MT score and thought we were talking threadripper. That's the absolute state of corelets, they simply can't believe this. It's a sad state of denial.
>>
File: bk.jpg (354 KB, 921x865)
354 KB
354 KB JPG
>>65557533
JUST!
>>
Holy shit, I was expecting just a simple die shrink, not such gains.
Imagine Zen2 on 7nm.
>>
So zen2 potentially will be 4.8ghz stock ?
>>
>>65557464
MAN I HAD IT WITH THIS FUCKING JEWS!!!!
I AM SELLING MY SHIT i7 6700k!!!!!
FUCK IT!!!!
>>
File: 1518073993045.png (215 KB, 432x616)
215 KB
215 KB PNG
is zen still ass on linux?
i just wanna build a comfy server
>>
>>65557633
That board is probably the best for water cooling 8700Ks even better than most Z370s
>>
>>65557648
No. Just make sure you have an up to date kernel and BIOS.
>>
>>65557648
Still? Agesa updates came out like, 2 months after zen launch.
>>
>>65554128
No, their current 14+++++++nm process lets them jack their clocks high enough to surpass the 2700X and a even higher clocked 2800X
Even without the higher clocks Intel still beats AMD at extremely niche and really extremely latency sensitive workloads, like HFT jew nigger shit
Intel won't be finished on every front until 7nm, unless they get their 10nm process up to speed or some other miracle happens
>>
>>65554669
That's what happens when everything has been optimized heavily for the stagnant Intel arch's since Sandy Bridge, this optimization is quite significant when you compare it against a completely brand new CPU with zero optimization
>>
>>65554696
>>65556632
HT doesn't really suck, it's just that Zen is able to execute more instructions at once than Intel
Since it's really hard to extract so much parallelism from most programs, running two at once really helps to make use of Zen's execution units
>>
File: ult.jpg (11 KB, 480x360)
11 KB
11 KB JPG
Great thread as always guys
>>
>>65557933
You don't execute 2 instructions at same time it's impossible with ht or smt. You just have a pipeline with different locks waiting resources our running on CPU.
>>
>>65557592
>Much gains
It's basically mediocre gains. Anything's for such shitty frequency is a gain
>>
>>65554128
> 8 core matches 10 core in multithread

CAN SOME ONE EXPLAIN TO ME WHAT THE FUCK IS GOING ON HERE ?
>>
>>65557972
lower clocks and not running the mesh at 3200 or higher
>>
>>65557955
Processors have been superscalar for a long time now. I think even the pentium1 was superscalar. Multiple integer execution units can execute multiple instructions at the same time if the processor knows there will be no side effects from doing so and that the results do not rely on each other.
>>
This is what Ryzen 1 should have been. But I'm still going to wait until 7nm Zen, only then Intel will be truly fucked
>>
>>65558113
>Intel will be truly fucked
Doubt it.
>>
>>65557972
this is sorcery, does Lisa Su acquired some alien tech?
>>
>>65558204
Jim Keller is clearly not from this world
>>
>>65558160
10nm will never happen. They'll just rebadge last year's CPUs with a tweaked 14+++++++++++++++++++++++++++++++++++++++nm, 10GHz clocks and a 300W TDP
>>
>>65557385
>absurd overclock
>not showing cpu-z's benchmark
>>
>>65557972

Something something, Zen actually can run more (specific I think?) instructions per clock cycle, per core(don't quote me on this) but is bottlenecked with one thread on one core, but not with two threads leading to the almost linear scaling of Zens multi threading. Really I don't remember exactly, there's some good technical stuff on Anandtech about it from a user called "The Stilt". Posting from phone so fuck you find it yourself it's a interesting read if you like that stuff.
>>
>>65554224
>>65554243
>>65556950
>"turbo"/"boost"
I really don't like how both AMD and Intel manage to fool tech press and others into showing "turbo" speeds. It's pointless and useless. My CPU is either close to idle or all cores are working at or near 100% load. "turbo" and "boost" (whatever they call it) only matters in synthetic benchmarks. Just tell me the IPC or each thread when threads are maxed out.
>>
>>65554128
Now I might upgrade from my 4790k
>>
>>65558350
Normies usually don't actually use their computers much, they either only do short spikes of load while loading facebook and scrolling through their feed or medium loads with a few spikes while gaming
>>
>>65557912
Thing about that is, for the most part code that's friendly to Intel is friendly to ryzen. But old software using old compilers uses unoptimal code paths treating it as older AMD CPUs. That's where in titles such as Rise of the Tomb raider when ryzen seen a significant performance uplift, this was mostly a simple code path update and a few tweaks. Some of those tweaks also helps Intel cpus as well.

You can look at Prime95 as a perfect example of this. Anything before 29.1 couldn't get ryzen beyond 80% load due to using unoptimzed code since the software had no idea what it was looking at, or what to do with it.

>>65557981
7900x as a 4ghz all core turbo, and can boost 2 cores up to 4.5ghz. Even if that 2700x is overclocked to 4.3ghz all cores, there's not a big enough clock speed difference to compensate for having 25% less cores, and single core is still at a disadvantage. It's mostly down to mesh and the 7900x's pitiful cache.
>>
>shitposting amdrones still rampant
why can't they just wait for actual benchmarks that release in 3 days? what's the point in shitting up the board? if ryzen is so good then you should just wait for benchmarks.
>>
>>65558433
>inb4 actual benchmarks are even better
>>
>>65556515
IPC has always been at this level. Ryzen was just lagging in clock speeds
>>
>>65556603
Not unless Intel's 10nm process is successful. Because last I checked it was a dumpster fire.
>>
>>65554333
delid this
>>
I wonder how far you could push the 2600 safely
>>
>>65559010
probably ~4.2ghz if you get a good one or ~4ghz if you get a bad one.
>>
File: Untitled.jpg (198 KB, 1148x398)
198 KB
198 KB JPG
>>65556950
Something is wrong with your haswell.
>>
>>65559010
enough to make it a very solid mainstream option.
>>
>>65559078
he's clocked 600mhz lower than you and I assume also using DDR3-1600 rather than 2133 or 2400.
>>
File: Lisa Su 1.png (46 KB, 150x262)
46 KB
46 KB PNG
How big is Lisa's strap-on?
>>
>>65559094
I do have 2400Mhz ram so you may be right. His single thread speed still seems kind of weak.
>>
File: intel-crematorium.jpg (315 KB, 800x531)
315 KB
315 KB JPG
>>65554128
>All these Intel wojaks
top kek thanks for the laughs lads
>>
File: 1050$builds.png (256 KB, 1900x780)
256 KB
256 KB PNG
Not shilling here but the ryzen is probably clocked at 4.4 or 4.5ghz all cores
I dont see it beating skylake or newer intel in IPC yet.
Expect a 10%~ gain in overall performance over a 1800x

Reminder of the type of build you do for 1k~ when ryzen was first launched
>>
>>65554128
I would have preferred AMD to fuck Nvidia up in the GPU space rather than fuck up Intel in the CPU space
>>
>>65559212
that isn't going to happen anytime soon
>>
>>65559236
Navi sounds an interesting midrange GPU in the same general thinking as polaris, although that does mean little competition at the top end of the market again, which sucks.
>>
>>65554128
Threadripper 2.0 will be insane.
>>
File: 1520396382702.jpg (162 KB, 555x490)
162 KB
162 KB JPG
7nm is going to be fucking stupid fast
>>
>>65558613
You don't need new processes for new uarches.
>>
You should wait until the benchmark is fixed before posting these obviously fake results
>>
File: 1510882587301.jpg (2 KB, 125x100)
2 KB
2 KB JPG
>tfw i'm going to buy this to replace my 1700x before the reviews come out because i'm an impatient idiot


send help
>>
>>65559431
don't be a Muppet lad, you have a really nice little cpu there, you're just throwing your money away.
>>
>>65555385
if they could release it and legitimately blow intel the fuck out they would have done.
>>
>>65559458
but it's better !
>>
>>65559431
You're covered for a good 5 years with that CPU, why upgrade?
>>
>>65559557
it looks a decent amount faster with lower mem latency and better support for higher frequency ram
>>
>>65559431
That's a very nice CPU you have there.

Wait for Zen3 or even later if it will still fit in AM4. It will blow your balls away.
>>
>>65554128
dont worry anon cpuz will release a fix to fix the numbers of the new ryzen+
>>
5.8GHZ
>5.8GHZ
5.8GHZ
>5.8GHZ
5.8GHZ
>5.8GHZ

https://twitter.com/anthony256/status/985784728258097152
>>
The real question is - should I finally replace my i5-2500K with this?
>>
>>65559877
yes
>>
>>65559877
do it
>>
>>65554696
Intels hyperthreading sucks, amd has a better implementation
>>
>>65557933
So basically AMD is better in most programs and Intel is selling you retards memes glued together with pigeon shit. Amazing.
>>
>>65554128
*generic big if true reply*
*followed by delete this*
>>
File: unknown.png (240 KB, 1394x1268)
240 KB
240 KB PNG
>>65556950
That was already the case with Zen 1, but now with more clock and optimization.
>>
IPC stands for instructions per cycle not core you fucks
>>
>>65560718
k
>>
fuck off squint
>>
>>65559758
1.76v and 56% OC over stock means it'll be pulling 105W * 4.8x ~ 500W
>>
>>65554128
No, this means intel is going to be revived, instead of just being apple's bitch that keeps just increasing the igpu performance.
>>
>>65554128
more screenshots? like RAM speed and Clockspeed?
>>
File: 1600.png (16 KB, 403x402)
16 KB
16 KB PNG
>>65554128
Around a 10% single thread improvement compared to my 1600 at 3.95GHz.
>>
>>65556632

HT is just Intel's marketing term for SMT.

It is just their chips since Pentium 4 had resource contention issues. They didn't bother trying to fix it until Skylake-X.
>>
>>65557421

It is because that Z170 board is hilariously over-engineered. Its older Socket 1151 layout is able to deliver enough power to keep Coffee Lake silicon happy.

The problem is that majority of older Socket 1151 boards (Z1xx/Z2xx) board aren't like this. They would have trouble powering fully-enabled Coffee Lake silicon without stability issues. You have to resort to hard-modding to get it working as intended or underclock/undervolt it. Intel wanted to avoid a repeat of PIII 1.13Ghz debacle with its OEM partners.

Intel never intended on making six-core 14nm SKUs for Socket 1151 platform when it was originally designed. Coffee Lake is literately a rushed, last-minute change because of Ryzen 7 family.
>>
>>65559877
I'm going to do it, personally
>>
>>65561642
Ryzen 2700X or 8700K or that 8 core?
>>
>>65561676
>Ryzen 2700X or 8700K or that 8 core?
well, wait at least for Ryzen+ review in a few days. Intel will launch their own mainstream desktop 8 core on Z390 chipset but i dont know when. check for yourself and buy what suits you
>>
>>65554364
Hnnnnnmg I want Ryzen 7nm nao
>>
>>65555385
Maybe it'll be highly binned dies that can do 4.5GHz+
>>
>>65561206
Yes, at less than 10% higher clocks. Just as expected it has a slight IPC boost with it's higher clocks. What's your point?
>>
>>65562124
Nah, AMD is probably binning them for Threadripper+ and Epyc+. We don't know what the yields for 12nm are especially in comparison to the existing Ryzen fabbing.
>>
>>65554128
Why is the Shit Wreckers touch so effective?
>>
>>65562164
>Epyc+
there will be no such thing, TR+ and Ryzen+ only. EBYN will get 7nm treatment sometime next year
>>
>>65562246
Well, in that case the super high end binned dies are likely being saved for Threadripper+
>>
>>65562263
guess so
>>
>>65562246
There's no Ryzen+... even Lisa Su uses Ryzen2 /Ryzen 2000 on twitter posts.
>>
>>65557583
is this fucking real??!?!? ahahahahaha
>>
File: 1508108137217.png (986 KB, 1400x1000)
986 KB
986 KB PNG
>>65554128
>>
>>65562161
There was no point, you defensive, AMDrone dipshit. It was just providing a point of comparison.
>>
>>65557626
For 2 core boost? Probably close to that. All core turbo? no.

>>65557933
HT and SMT don't "execute two instructions at once".
They allow the core to be used while it's otherwised locked.
AMD's SMT is simply much much efficient at reacting to this locking and allowing another thread through.

>>65557972
It's not exactly new? 6 core Zen1 almost matched 8 core broadwell-E.
AMD's SMT is just so much better than HT. It gets over 50% scaling in some cases (so 8 cores are effectively 12), while the best Intel manages is 20-30%. In CPU-Z, 8 cores are like 10.7 for Ryzen while 10 cores are like 11.28 for Skylake-X.
>Intel has 11.28% HT scaling while Ryzen has 33.75% on CPU-Z.
It's also more efficient so the all-core clocks tend to be a bit higher for a given TDP than Intel's advertised single/dual core speeds.
Hilariously, the 2700X is a 105W part while the i9-7900X is 140W.

>>65562316
Ryzen2 = 2000 series.
Ryzen2 != Zen2.
>>
Can't wait for the anger and denial of the amdrones when the cpus and benchmarks of credible sources will come out. It happened before, it will happen again. You can post crying wojaks in reply now, thanks.
>>
File: 1477150621579.png (42 KB, 653x726)
42 KB
42 KB PNG
>>65563038
Gotcha senpai
>>
>>65562965
What makes you think Zen 2 won't hit 5Ghz? Globalfoundries are targeting 5Ghz for 7nm according to their slides.
>>
You can tell when the hardcore AMD dipshit is around when he starts repeating IPC in every post.
>>
>>65563038
In all fairness AMD probably has something actually competitive this time.

Its just too bad that the rumor mill has already moved on to the new Intel leaks.
>>
>>65554333
>$999
>>
File: 1521352514969.jpg (218 KB, 685x566)
218 KB
218 KB JPG
>>65563312
Too bad intel does not solder the IHS , that would fuck AMD even more by not having to delid the CPU and risking it to get fucked more people would buy the high end stuff, but noooo lets save us some shekels and use jizz on them...also AMD is more compeling for those Twitch guys than anything, server side the might the edge because of meltdown...but anyway competition only benefits us customers so...let them fight..
>>
File: 1523448103584.jpg (511 KB, 2000x1333)
511 KB
511 KB JPG
When will we see the review embargo end? Cant wait to laugh at amdrones for over hyping it, at which point i will be told it was never supposed to me good.
>>
>>65559431
Why? It's not a huge difference, just overclock it and wait for Zen2/3.
>>
>>65563440
19th of april 2k+18.
>>
>>65563417
They way they do it now makes their chips more durable, it isn't a disadvantage because they already stomp AMD.
>>
>>65563038
>credible sources
I bet you need a certified kike employee to double-click cinebench.exe
>>
>>65554669
It wasn't just a piece of code it was good at, it was just smart enough to skip over a useless pause essentially. Since it skipped over what was an intentional pause, it was arguably breaking the rules of the benchmark.
t. 1700X owner
>>
>>65563288
I think it won't be 5Ghz STOCK nor on ALL CORES. I think most chips will hit that on OC.
I just don't have a lot of faith in a brand new process with no engineering samples. 7nm+ sure.

Same way 14nm LPP was "up to 4GHz" while all core clocks were like 3.7GHz on the top end stock.
>>
>>65563520
How does using shit thermal compound make it more durable? Would using better one not lover temperatures and thus make it serve longer?
>>
File: t.JUSTnich.png (1.25 MB, 1002x1480)
1.25 MB
1.25 MB PNG
>>65563520
>They way they do it now makes their chips more durable
stop shitposting brian, it's to worth it
>>
>>65563560
The rare ones do 4.2+

https://www.youtube.com/watch?v=VSbIN59uUOk
>>
what are the chances motherboard manufacturers won't abandon x370 bios?
pretty low?
>>
>tfw Lisa Su cucks your gf by raping ur bumhole via strapon

REEEEEEEEEE
>>
>>65555816
All their CPUs come with a cooler now though. So I doubt that.
>>
chipset doesn't do anything performance wise for ryzen, like at all
why would it run slower besides raw BIOS?
>>
>>65563633
Yes. I've even seen threadripper at 4.3GHz all core at 1.35v stable in stress testing. But those are rare and overclocks. The all-core turbo stock is 3.7GHz.
I do think 7nm will likely hit 5GHz or close to it with some samples overclocked. I don't think they'll be stock clocks that close to it until 7nm+ though.
SORRY FOR THE REASONABLE POST.
>>
>>65559095
>that pic
The bottom should be changed to
And basically Intel is BTFO!!!
>>
Yearly reminder that before Ryzen family of processors launched, octocore cpu's went for upwards of 1000 dollars and it was a hilarious pipe dream to even think of a 500 dollars or under 8 core cpu.
When the first rumors for the Ryzen began to spread in the wild, there was much skepticism that an 8 core cpu would cost anything below 800 dollars. You can easily find a couple of youtubers discussing exactly this and how Ryzen would fail because they were absolutely sure thhat these processors were going to be ultra high end, in terms of price point.
The 1700 launched at 329 usd msrp
This was a very big surprise
there's a sale right now for a 229 Ryzen 1700 at microcenter

I just want to make intel shills cry moar, not gonna talk benchmarks. But I want to make clear that Ryzen changed the short term and quite possibly the long term prognosis for the consumer processors for good. And that's not even debatable.
>>
>>65564005
>Yearly reminder that before Ryzen family of processors launched, octocore cpu's went for upwards of 1000 dollars and it was a hilarious pipe dream to even think of a 500 dollars or under 8 core cpu.
Why are you forgetting the FX-8370 or FX9590, these may have sucked with IPC but trashed all of intels offerings in multi-threaded applications.

Oh they had 8 cores
>>
>>65564005
There was also just as much speculation that it would land at 8cores for 300 to 500 dollars and it did both. Was was probably the idf saying 800dollars trying to draw negative attention.

Anyway, it was cheap for the 1700 and the old "x" skus have been revamped as they weren't as good value. The 2600x for example lost all its good value without a cooler, which was just some shit they copied from Intel like phone manufacturers copy apple.
>>
>>65564043
8 integer cores
>>
>>65564043
you know exactly what I meant.
>>
>>65564043
You'll be hard pressed to find a bigger unironic AMD fanboy in this board then myself. With that said: Bulldozer was shit. It's history now.
>>
>>65563588
if its more durable then the consumer won't have to replace it after 2-3 years. i dont think you will make it at Intel sorry
>>
>>65564043
Not it didn't, fuck off with your history revisionist bullshit.
>>
File: 1523031906083.jpg (116 KB, 1280x720)
116 KB
116 KB JPG
>>65554128
Intel is going to be blown the fuck out again. Only a few more days left.
>>
>>65559269
Well mid ranged is a far bigger market. Like 60-70% so its certainly the more feasible for them. They just need to get their feet on solid ground cause its been tough for them for more than a decade
>>
>>65554364
> seems to be only 2.5% away from a i7-8700k at the same clock speed

Wait isn't the 2800x a direct competitor to the coffeelake 8700k?
>>
>>65564043
the FX 9590 launched for 1k, though.
>>
>>65564213
We all know halo effect is THE truth for the gpu market. As anedoctal evidence, I have many acquaintances who have always fawned over the 1080ti's but due to budget constraints got 1050ti's or 1060's, including many 3gb models, wihout a single pause for thought and comparisons. There are a bit more then a dozen people in this informal sample size.
>>
>>65564186
it would be literally anudda shoa if this pic isn't shopped
would love to know what RAM and Clocks were used
>>
>>65564123
History has shown that bd based opterons were the best choice for gpgpu computing.
>>
File: 1522778603206.jpg (5 KB, 168x168)
5 KB
5 KB JPG
>>65564333
checked
>>
>>65564043
lol no, the 4790k humiliated AMD so hard that it wasn't not even fun
>>
>>65561182
You do realize this competition helps all of us. AMD is competitive on price because at that price point you have to shell out double for an equivalent intel series. Plus every generation you need a new mobo whereas AMD let you use the same motherboard for future upgrades. For a budget you get superior performance.

Remember before ryzen, high end multi-cores cost upwards of a $1000. Also they long standing security flaw with their chipsets taking forever to solve.
>>
>>65564140
Doesn't high temperatures slowly fuck up cpu? I do not see how letting cpu get hotter makes it more durable.
>>
>>65564364
>security flaw with their chipsets taking forever to solve
they halted further patches.
>takes forever to solve
this checks out. When you pause the effort to fix one problem the time-to-solve tends to infinity. as another way of putting it.
>>
>>65563823
Still water cooling with this setup?
>>
File: 56745667.jpg (7 KB, 200x213)
7 KB
7 KB JPG
>>65554128
>2018
>using CPU-Z
>>
Oh boy cant wait for new threadrippers. Gonna get one
>>
>94C is fine
This is your brain on intel
>>
>>65564631
he doesn't understand the intrincasies of a leak.
he never thought about the fact that one benchmark has its merit not for the purpose of being thorough but due to its popularity, no matter how shitty it is.
>>
>>65558350
Intel doesn't want you to see the IPC, or you might realize that they've only been coasting on superior process all this time.

Ryzen is the superior architecture, clock for clock. And it's fabbed on some shit tier process, so you can barely reach 4Ghz.

Good news is GloFo/IBM are catching up, what with Intel completely fucking up 10nm. The reason Intel is so scared releasing 10nm, is that since the process isn't mature and their IPC hasn't improved in 5 years, anything they release will just be a clear regression.

When GloFo 7nm comes around next year it's going to be a fucking blood bath in the Intel investors meetings.
>>
>>65554128
>beats the 3.3ghz processors in single thread
Are we excited about this? Unlocked i5s do over 5ghz
>>
Only disappointment for me in this new lineup is that they gimped the turbo on 2600X vs 2700X.

1600X and 1800X had same clocks.

AMD, why can't you same frequency, lower core count?

Also, because it's a refresh, a lot of people interested in it already have a cooling solution.
Why you bundle a cooler in everyone of them 2000 cpus, while you didn't do that with Zen 1?
>>
>>65564816
>Unlocked i5s do over 5ghz
That's not a good thing for Intel.

If AMD CPUs are competitive with CPUs that run way higher frequencies like the i5s, that literally means they have much better IPC.
>>
>>65564846
>AMD, why can't you same frequency, lower core count?
Because lower binning.

Good news is if you have less cores, you automatically consume less power, so you can go crazy on the voltages and overclocking.
Reach the same frequency easily even on a cheap board.
>>
Drones should just go back to /v/
>>
>>65564847
The 8400 @ 3.8ghz has higher singlethread performance than the 1600 at 3.8ghz. We will see what the 2600 does but i imagine it will still be less
>>
>>65564846

>Why you bundle a cooler in everyone of them 2000 cpus, while you didn't do that with Zen 1?

Wait, I found the answer myself.
It's so that I can sell my used 1600X with a cooler, and get an almost free upgrade.
Thanks AMD.
>>
>>65564914
So the current situation is if you have a workload that's not frontend-bound, memory/cache bound and not AVX2, Zen is actually ahead of Coffee Lake in IPC.
Unfortunately for AMD, that does not include most benchmarks and games.

The biggest IPC problem with the current Zen architecture is the ifetch bottleneck, and the caches being plain awful.

The caches are already bad at latency, and they only have 32 bytes/cycle read bandwidth instead of 64 B for Intel.

The Zen backend on the other hand is absolutely overkill, ignoring AVX2 it's solidly ahead of the Intel backend.
This makes hyperthreading very viable on Ryzen, there's just way more than enough resources on the backend for two threads to share.

Now if AMD would just fix their shitty caches and beef up the instruction fetch just a little bit, they could give Intel a real run for their money.
>>
IT'S UP!

https://www.youtube.com/watch?v=Q5GoHmtnmus&t=0s

https://www.youtube.com/watch?v=1Cxfl9dkeqU&t=0s
>>
When are the iGPU versions coming out? Need to up my shitbox soon
>>
>>65565133
The Ryzen APUs are already out.
>>
>>65564394
>if your gonna do something. Do it right
>>65564326
Yeah that's just the fact if the matter. Majority if the people just dont have the money or can justify spending $700+ on one computer component
>>
>>65559431
Please go ahead. I could go for a cheap barely used CPU for my 5 year old system upgrade.
>>
>>65565121
The delid is post binning for SL
>>
Want best multithreading, go AMD
Want best 1080p gaming, go Intel
Want best >1080p gaming, go AMD, and use saved $ to buy better graphics card.

There you have it, the only people defencing Intel nowadays are 1080p 144Hz gaymers.
They most likely fell for the G-Sync meme too.
>>
>>65565453
The binning is voids guarantee.
The overclock is voids guarantee.

Does nobody consider guarantee as important anymore?
>>
>>65565348
If you're in yurop, I can sends you a never overclocked 1600X.
If you're patient, I can sell you the 2600X cooler with it.
>>
>>65565674
I'll pay the shipping cost for the cooler and €130 for the 1600
>>
>>65565611
I have a 1700x a 1080ti and a 144hz 1080p monitor. Literally what game am i not getting the full benefit of my 144hz game? PUBG? Literally 10-15fps behind intel in worst case scenarios.
>>
>>65555981

He was implying those were not taken into consideration, not actually asking.
>>
>>65565710
Nigga jacking my shit
>>65565674
Its funny you say that. Within a month I'll have moved to wales but I'll need to settle down before I can buy anything so it'll take me about 3 months or so to try buy it off you
>>
>>65554364
>In games. Only gamres are important.

t. intel rep
>>
>>65554696

I7-7900x is the retarded Skylake X processor. That was true even before Ryzen. Let’s see how it does against a 7820x
>>
>>65565710
I would have said 150 for both, but it really depends where you live, I guess.
Anyways, I'm stupid, it's not like it's wise to trade anything around here.
Just look at e-bay or similar in the next few weeks, and I'm sure a lot of people will be puzzled with wtf they're supposed to do with that new cooler.
>>
>'But they did not test with an overclocked GTX 1080 Ti!'
>Implying anyone can afford to buy one on this board
>Implying anyone who does represents the majority of gamers
>Implying it would be any different if there were no crypto niggers
>Implying I give a shit about JUSTlel anymore
>>
>>65564550
The reason they did not solve it is because it impacts performance for certain server applications by up to 34%. Theye did not want a performance hit that would put them at a disadvantage. So they tried to bury it as if it was not really that important a thing to worry about.
>>
>>65554198
>>only 15.6 more in single core performance against an old gen at 3.3GHz/4.30
>old gen
>single core
You know that there's no ipc difference between 7th and 8th gen? 8th gen is literally 7th genwith extra cores GluedTogether™
Your employer "forgot" to convey this fact to you it seems.
>>
>>65565719
Well, it's just that you don't play CPU intensive games.
Happens to everyone.

But buying a 1080Ti to play 1080p?
I'm sure it's doing nothing, most of the time.
Must be quiet, though.
>>
>>65565763

>I7-7900x is the retarded Skylake X processor.

Retarded how? Is the 7820 objectively the better one? I was thinking about buying a 7820x but now it seems like i should just wait or get an 8700k for less money
>>
>>65565879
Tell me which game where getting 144fps+ is important where my ryzen would be insufficient.
>>
>>65565982
Starcraft 2
>>
>>65565968
Imagine possessing a 7740X config, this instant.
>>
File: 4qsfhkk.png (129 KB, 1562x332)
129 KB
129 KB PNG
>>65563560
>Same way 14nm LPP was "up to 4GHz"
Actually, it was "greater than 3GHz". On desktop Ryzen we got base clocks ranging from 3.1 on the 1200 to 3.8 on the 1900X, easily meeting that goal.
>>
>>65566107
That 5GHz 7mn makes me fucking hard
>>
>>65565993
Wrong. Starcraft 2 is about APM. Frames literally don't matter as long as you have enough frames so that input is never dropped
>>
>>65566045

It's really that bad? Everything about it makes it seem like the 7900x would be better
>>
>>65561167

So equivalent to a Skylake-X running in power saving mode then.
>>
>>65566178
4 cores/8threads in <current year>
112W tdp.
Higher priced than 2700X.
Was the 'response to Ryzen' before Cofee Lake.

So idiots actually bought it.
>>
>>65558160
I seriously wonder what they could offer that could deal with AMD's 7nm, 5Ghz target (base clock) product.
They have nothing that is even technologically capable of similar things, especially at higher core counts.
>>
>>65566230
Wait for sekrit sauce lul
>>
>>65566228

I was talking about 7820x vs 7900x. I already knew the 7740x was pants on head retarded.
>>
>>65566283
Well, desu, The whole Skylake-X lineup is retarded.
I would be viable if prices dropped twice-fold.
>>
>>65566230
AVX4 ( 1024? )
>>
File: 1246602885370.jpg (72 KB, 460x460)
72 KB
72 KB JPG
All these battles of inches on /g/ is really fucking stupid. Be an adult, purchase whatever is good enough to meet your needs,and has the best performance for your dollar at that category. When you're shitposting all day the only difference you're gonna feel is how much extra money you're sitting on.
>>
>>65566376
No way, you are being too rational. Expect getting called jew/uma delicia or pajeet by 12 year olds.
>>
>>65565121
Holy crap did he just fucking dis AMD fanboys!

https://youtu.be/1Cxfl9dkeqU?t=1097

He ripped you all a new asshole!
>>
>>65566551
All fanboys are retarded
>>
>>65566376
>be an adult
Nigger do you seriously believe that the percentage of adults here is more than a single digit?
>>
>>65566584
>You will immediately cease and not continue to access the site if you are under the age of 18.

Considering the rules as stated above, I hope so.
>>
>>65566551
wrong pasta

>>65566589
>>
File: Threadripper 2 leaked.jpg (128 KB, 500x332)
128 KB
128 KB JPG
have a leaked shot of 2800x
>>
File: 1521686842896.gif (91 KB, 200x200)
91 KB
91 KB GIF
>>65566863
>>
>>65566623
lol
>>
>>65564755
I so wish you're right, anon.
>>
>>65554198
Turbo to 4.5, more RAM channels and lower latency, monolithic.
>>
>>65565133
Like a month ago, r3 2200g and r5 2400g
>>
>>65562359
aye brother
https://twitter.com/bkrunner/status/940795579491381249?lang=en
that first comment though
>>
from being sensible for a eternal Intel tripfag like this >>65563312 then go full 180 degree >>65563520

>thin substrate
>mayo thermal compound
>makes chips durable
jesus christ
>>
>~360 ST, R5 1400 stock
>~1920 MT, R5 1400 stock
Hmm, I wonder how far overclocking can get this...
>>
>>65571095
>didn't get 1600
cuck
>>
lmao intelkikes on suicide watch
>>
>>65571382
>caring about which CPU somebody else bought
When I bought it, the 1400 was $149 AUD. The 1600 at the time was $295. Same shop. Fuck Australia.
Paying almost double the money makes no sense, for only 50% extra cores and threads.
>>
>>65562965
>HT and SMT don't "execute two instructions at once".
>They allow the core to be used while it's otherwised locked.
>AMD's SMT is simply much much
efficient at reacting to this locking and allowing another thread through.
That's not what it does, it allows to execute instructions from another thread in the same core, this helps a lot on Zen since it can do a maximum throughout of 6 instructions per cycle, unlike Intel's 4
SMT is used since getting more than 2 or 3 instructions that don't depend on each other is really fucking hard on almost any workload, without SMT execution units go unused most of the time
Read up on superscalarity and out of order execution
>>
>>65564816
How delusional are you? 5+ GHz 8700K are pretty rare, it skyrockets with lower bins into unicorn rare
>>
>>65571868
If your chip can't do 54/50 it's a dud.
>>
>>65565121
you people should watch until the end
>>
>>65565650
>The overclock is voids guarantee.
Does it really? In Europe?
>>
File: 1519229607315.png (400 KB, 816x1020)
400 KB
400 KB PNG
>>65557972

THANKS JIM
>>
>>65554128
Daily fucking reminder, to all the newfags in this thread:

IGNORE the tripfaggot
DO NOT respond to the tripfaggot's posts
HIDE OR FILTER the tripfaggot
>>
File: 1523730887356.jpg (31 KB, 470x470)
31 KB
31 KB JPG
>>65566186
>>
>>65554198
>300$ processor
>15.6 more in single core performance against a 900$ cpu
>>
File: 1506383390040.png (614 KB, 992x1043)
614 KB
614 KB PNG
>>65571603
MORE NIGGAHURTZ
>>
>>65571849
You look it up.
https://www.ibm.com/support/knowledgecenter/en/SSB27U_6.4.0/com.ibm.zvm.v640.hcpa0/mtfac.htm

SMT implementations of IBM, Intel, and AMD allow two threads to share a core. That means they can't use up all the execution units at once, including the scheduler. This primarily means one thread using a core while another thread is in a wait/lock state.
>>
>>65573999
CHECKED JIM
>>
>>65571796
It is worth it in the long run. Also why didn't u buy one with the 20% off eBay deal shit. Ez 1600x for less than 300 aud
>>
>>65564917
I remember that they were going to release versions with an RGB variant of the wraith cooler, but the jews didn't like the idea of a CPU including a good stock cooler so they threatened to not make any AM4 cooling solutions if they released those.
>>
>>65575980
The wraith spire on the 1700 was rgb, or at least red led, I didn't change it.

It was the wraith max (prism is probably the same spec) that oems apparently complained about. But amd doesn't care about that anymore. The x editions were poor value because they didn't include a cooler whatsoever. I guess they should just bundle the spire cooler for everything and have them only be adequate for the x models and if people want more they get better coolers as usual. That'll avoid the stupid having four different coolers across four skus.
>>
>Every time I OC my 1700x my computer crashes after approx. 1 hour
>Have to stay at base clocks if I want a stable machine
>CPU doesn't know it can boost by itself because no games ever stress it properly so it just stays at 3,4 100% of the time

I wish I knew what the problem was, I had it running at 3,8ghz last time I tried but then it just crashed. When I got it I had it at 3,9 for a while but it wasn't stable
>>
>>65576674
Check your thermals, voltage and for bios updates.
>>
>>65576696
Thermals are below 60 degrees (celsius) all the time thanks to my NH-D15
Voltage defaults to 1,35, might be too much for just acheving 3,8, think 1,3 might do it?
And bios is the latest version
>>
>>65576674
incrementally increase your clocks start at like 3.5/3.6 stress test for around an hour then bump it up 100 Mhz. Use either P95 or AIDA64
>>
>>65576674

Spend more than :10bux: on your psu next time.
>>
>>65557385
>im lucky and had enough autism to get a hacked bios that means eveyone is and does
mong
>>
>>65557533
He could potentially not be a shill. After years of sucking intel cock, he probably autoatically thinks 1000+$ when he sees 8 cores
>>
>>65573999
JIM KELLER
JEM KILLER
JEW KILLER
>>
>>65578667
You forgot your trip. >>65563520
>>
>>65559877
no
>>
>>65559877
yes
>>
>>65559877
maybe
>>
>>65559877
perhaps
>>
>>65578780
>>65578962
>>65579198
>>65579201
KYS
>>
>>65560718
IPC stands for inter port communication
>>
>>65579500
no
>>
>>65579500
maybe
>>
God I hope 7nm zen will be good. If it is I'll definitely upgrade my 3570k finally.

I could upgrade now, but I want massive improvements because I have to basically buy a whole new system anyway.
>>
>>65565762
Not anymore goyim, its all about ps3 emulation now or whatever incredibly niche shit intel is still good at
>>
>>65579863
>or whatever incredibly niche shit intel is still good at
lel, expect them to shill AVX2 even harder

oh wait cheaper Intel CPUs either don't have it or it's crippled to the point of being useless
>>
>>65579885
AVX2 GOYIM, AMD BTFO, I USE AVX 24/7
>>
>>65579885
Intel's advantage with most AVX isn't even that large due to how much slower they get clocked when running it, and their core/price disadvantage. And afaik, Zen3 is surely to have AVX512 support.
>>
>>65573999
THE JEW DESTROYER
>>
>>65580317
shut up goy
>>
>>65579863
240p gaymen benchmarks.
>>
>>65579625
>>65579776
KYS
Y
S
>>
>>65580657
no u
>>
>>65580175

>Intel's advantage with most AVX isn't even that large due to how much slower they get clocked when running it,

Serve the home disagrees.




Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.