Intel Core i9-7960X performance; lands a 3200 on Cinebench R15.

>>61789142
being a thousand dollars cheaper

>>61789142
By not requiring delidding with a phase change cooler to avoid throttling.

>>61789177
>As if that matters when you are the king of Cinebench

>>61789142
Not shown: 2KW PSU to run the CPU and several gallons of LN2, local fire department was lead to the scene shortly after benchmark

TDP: 750W

>>61789142
they have an army of rabid fans but they're all poorfags

>>61789142
How much does it cost? $2000?

>>61789142
the TR 1950X scores 3277 in CB R15 , so it beats this thing , thats how they compete
oh and its half the price and has 64pcie lanes / quad mem

File: Capture.png (37KB, 1254x429px) Image search: [Google]

37KB, 1254x429px

>>61789142
what do they mean by this?

>>61789240
>engadget

pure fucking trash

>>61789254
it's from OP's link.

File: cheeto.jpg (110KB, 497x640px) Image search: [Google]

110KB, 497x640px

I don't even care who is actually faster AT THIS VERY SECOND.

It's just so damn refreshing to witness a real race between those two again.
Reminds me of the old early 00s megahertz race.

>>61789240
>consumer
>$2,000

*good goyim

>>61789177
The 16 core is actually $700 more expensive, the 18 core is $1000 more expensive.

AMD 16 core 3050 points $1000
Intel 16 core 3200 points $1700

So you get 5% more cinebunch points for 70% more price. And if you try and OC it, it will go into terminal meltdown. Whereas the AMD 16 core can OC to 3.9-4 GHz rather easily and get a similar cinebench score anyway.

>>61789142
Intel has 0 traffic on it's twitter, how is this plausible?

>>61789271
I'd rather not think about that because it reminds me of my friends who thought the turbo button made their computers faster

>>61789225
$1700. With fewer pcie lanes, higher power consumption, and constant thermal throttling.
Oh, and if you want raid you need to buy dongles for it. All this and the score isn't even 5% than the 1950X's.

File: tr amazon.jpg (219KB, 1536x2048px) Image search: [Google]

219KB, 1536x2048px

>>61789201
>damage control
>p-pls buy i9, pls

>>61789192
Threadripper is nearly a Ghz faster on it's base clock and won't require 600W

File: i9-7900X stock temps.png (53KB, 707x602px) Image search: [Google]

53KB, 707x602px

Threadly reminder that the i9-7900X can barely keep itself cool at stock with watercooling and the 12+ cores will have even higher temps.

>>61789407
no one runs p95 24/7 anyway spaz

>>61789421
What's the point in dropping 2000 grand on a CPU, if you can't use it to it's full potential.

>>61789421
Phew, good thing this isn't an enthusiast platform where overclocking is expected to work without causing a nuclear meltdown.

Oh, wait.


http://www.tomshardware.com/reviews/intel-core-i9-7900x-skylake-x,5092-11.html

http://www.tomshardware.com/reviews/-intel-skylake-x-overclocking-thermal-issues,5117-4.html

>>61789142
> Costs 70% more
> TDP is 50% more
> Fewer PCIe lanes
> Performs only 5% better
Intel is really dead, that 18-core will be the ultimate firehouse, and it will probably perform like shit due to shitty clockspeeds

File: 1502150543367.png (22KB, 1320x250px) Image search: [Google]

22KB, 1320x250px

>>61789718
3.4 GHz all core on the 18 core apparently. It's gonna be pretty damn housefire.

I'm predicting it will get ~3380 cinebrunch points. An astonishing 10% better. But again twice the price, insane temps, less PCIe lanes.

http://www.pcgamer.com/full-details-for-intels-core-i9-processor-lineup/

File: 1483167353491.jpg (23KB, 439x290px) Image search: [Google]

23KB, 439x290px

>>61789819
>3.4 GHz all core on the 18 core apparently.

>>61789271
damn now if we could get the nvidia vs ati times too it would be great too bad pajeet is a fuckup in the gpu department

File: 1497894370789.jpg (112KB, 691x771px) Image search: [Google]

112KB, 691x771px

>>61789177
THAT'S NOT FAIR

>>61789819
Get those compressors ready, we're up for some throttling!

>>61789142
>For a great VR experience
>VR

FUCKING KEK
YOU CANNOT MAKE THIS SHIT UP

>>61789906
Unless Intel grows a brain and uses solder, this is seriously going to be worse than the current crop of X299 housefires. It almost pains me to have to wait until October 18th to see the resulting damage.

>>61789142
the threadripper cinebench scores go to around 2900 at stock clocks.
The i9 7960X has 2 more cores, which should mean a 12.5% increase if IPC and frequency were equal.
3200 gives the i9 just a 10% advantage, and we know that skylake has slightly higher IPC.
This is for a full thousand dollars more, on a platform with significantly worse IO support.
Intel is shitting their pants violently rn.

File: AMD-Threadripper-1950X-Cinebench-R15.jpg (216KB, 1920x1080px) Image search: [Google]

216KB, 1920x1080px

>>61790181
>the threadripper cinebench scores go to around 2900 at stock clocks.
False. That score is from Dell's shitty ram. We already know the 1950x will do ~3050 at stock clocks and ~3250 at 4.0ghz all-core overclock.

>>61789142
I think I have a millionaire idea.
Put a store and sell pc hardware with payment/upgrade plans for +$1k hardware.

I will add insurance and pay you half the price for your hardware so you can upgrade one year after.

If apple could I can do it too

>>61789270
>>61789254
engadget's blogcrap on the i9 is 100% marketing paid for by intel. why else would intel's twitter twit about it?

>>61789852
>I am retarded and will keep repeating how Raja needs to leave
You niggers know nothing.

>>61789407
>mfw my friend just bought that processor

File: image_id_1882895.jpg (687KB, 3840x2160px) Image search: [Google]

687KB, 3840x2160px

found this on hwbot

>>61789283
actually if you oc the 1950X a little (aka locking all cores at 4ghz boost)
you get 3277 points and beat intel :P
but then again you can oc intel aswell , its just gonna run hot as hell and throttle mid CB run

>>61789271
Except they not only go up in performance they go up in price. This is not the 2000s i remember.

>>61789421
>no one runs p95 24/7 anyway spaz

It's not about running 24/7, it's the increased chances or crashing when stressed. If it's professional use it's a big fuck up, for gaymur money grabbing sure it's tippy top performance at the price of your kidney.

Intel used to be miles ahead with AMD just lagging behind.

Now Intel is getting BTFO by AMD on a scale of massive proportions.

Currently all Intel but if I will ever upgrade it will definitely be AMD for the price performance.

>How can Intel compete
>How can Amd compete
Make up your mind /g/

>>61789142
>>61789142
>>61789198
now i know why we had a fucking blackout in north east Africa

>>61789142
intel tweets engadget

i wonder how much they paid them for this

>>61789271
there wasnt a race back then only intel paying companies not to use the clearly superior amd

>>61790321
why is cinebench saying the x5650 is 12c/24t when its literally half that?

>>61792273
>Now Intel is getting BTFO by AMD on a scale of massive proportions.
only in marketing department
wait for quarterly reports

File: download.jpg (5KB, 225x225px) Image search: [Google]

5KB, 225x225px

>>61789142
Because you are going to need this to run an i9

>>61792702
dual cpu setup , says so in the details when you mouseover it

>>61792733
ah, makes sense. thanks

>>61789177

the 7700K is $100 cheaper than the 1800x and runs every game better at stock, much more at OC

you cant have it both ways

AMD is slower in everything, nobody is fucking transcoding 24/7 to care about the tertiary "benefits"

File: 1487459738216.png (6KB, 464x464px) Image search: [Google]

6KB, 464x464px

>>61794312
>1800X
>Not 1700
>muh gaymershit

File: file.png (2MB, 1920x1080px) Image search: [Google]

2MB, 1920x1080px

>>61789407

>18 shintel cores = 16 ayymd cores

rly made me think

>>61794492
what are you trying to say with that image?

>>61791129
You mean that person that you mooch support off? Because you're no friend of his, after allowing him to do that.

>>61794401
Try even an 1600x.

File: t.JUSTnich.png (1MB, 1002x1480px) Image search: [Google]

1MB, 1002x1480px

>>61794503

>>61794621

are you trying to say a 1600 can out perform the 7700k in OC gaming

>>61789311
thats literally what it did

>>61794715
No, just that saying a 1700 is comparable to a 7700k is crap, when the 1600x exists, and still kicks its arse in price/perf. The extra two cores don't play too much into it at this point.
Only retards to say that a 7700k is worth your time are in it to meme you out of money in six months time.

File: 1482448626332.jpg (25KB, 320x479px) Image search: [Google]

25KB, 320x479px

>>61789142
Fucking stop shilling this shit on /g/. You're cancer. I get that you're getting paid by Intel, but shit post in fucking moderation.

Nobody fucking cares. A thread died for this bullshit.

intel covers the spacex launch to send the machine on the vaccun of space to cool them down?

File: 1500694364786.png (371KB, 1277x718px) Image search: [Google]

371KB, 1277x718px

>>61794312
And the 1600 and 1600x are both over 100 dollars cheaper than the 7700k, on top of not being a furnaces stuck on a dead socket.

>Much more at OC
Loving every laugh.

>>61789495
>Tom's Shillware shitting on Intel for their bullshit and actually reporting on their glaring issues
Is this real life?

>>61794846
The 7700k takes the edge in gaming. If you're willing to have 80 rather than 87fps on your 60Hz monitor, then getting a much better processor overall is a better idea. For example; the 1600x.

>>61795092
You don't wanna be in mommy's blacklist.

>>61795092
It's when the cheques start bouncing

>>61795092
>>61795092
Intel has ran out of bribes money?

File: 1499347589150.jpg (58KB, 640x640px) Image search: [Google]

58KB, 640x640px

>>61795074
>reaching 90% 7700K
>memory not at or above 3200Mhz
that's damn good
are these with SMT off?

>>61795339
No, SMT on.

wait, kit WAS at 3200 >>61795339

>>61795353
someone should tell them to test off

>>61789240

7980XE is fucking "rejected" server-tier chip, overclocked.

It should have never existed. Not even workstation-types need that many threads on a single socket. 12-threads is about what most stop at before the thermals start to end way into the clockspeed/core count trade-off.

Server-fags will ignore it because it doesn't have ECC support. Any of the SP Epyc BTFO the 7980XE for about %40-50 less.

Only die-hard intel-fags with way more money then sense would get a 7980XE.

The only reason that anything above 7920X exists is because Intel marketing-fags want to keep up appearances in "core-count" game thermals and yields be damned.

>>61790321
oh so it's only performing 5% under despite having less cores and worse IPC.
[pant shitting intensifies]

File: 4ghz.png (589KB, 1609x859px) Image search: [Google]

589KB, 1609x859px

Barely beats the 1950X.

File: when-the-soup-is-not-delicious.png (350KB, 672x794px) Image search: [Google]

350KB, 672x794px

>>61789177
EXCLUA

>>61794312
The story is not that simple.
Games DO use more than 4 threads, but only when under stress.
Basically the intel CPUs increase the maximum frame rate, but the Ryzen parts increase the minimum frame rate.

>>61789142
> Twice the price
I can build a complete 16c system for the price of just the i9 processor
> Less PCIE lanes
> Resorting to launching your product on a gadget website to reach normies
> Guiding b.s writeups like this :
"During a routine business meeting at the company's Santa Clara headquarters, they decided to upend their desktop CPU roadmap for 2017 to prepare something new: the beastly 18-core i9-7980XE X-series.
Yeah because clearly this had nothing to do with your competitor drop kicking your roadmap and pricing tiers...
>It's the company's most powerful consumer processor ever, and it marks the first time Intel hsd been able to cram that many cores into a desktop CPU. At $2,000, it's the sort of thing hardware fanatics will salivate over, and regular consumers can only dream about."
It's a workstation processor..suited for workstation tasks. Learn how to market your product to the people you hope to win over.
>>61789282

>>61789283
They are on par which is why Intel didn't provide a live demo of them scoring that figure... Likely because they're using Turboboost 2.0 and 3.0 and it isn't easily repeatable.

>>61789294
mfw memetic energy isn't behind you

Gotta love the timing of intel's announcement... Days before AMD's product launch. Have some class guys.

>>61789819
Performance will be all over the place.. need a 4d chart just to explain runtime performance..
kek

>>61790428
>>61792612
You caught that too? They're sending one last call to the normies to rescue their platform. Gotta love normie sites... Always on the dole spreading bullshit.

>>61794503
kek

>>61795787
>that RAM

>>61794761
No it didn't. Turbo button down clocked the CPU.

>>61794312
>nobody is fucking transcoding 24/7 to care about the tertiary "benefits"
>im not doing anything other than playing games so no one else is either!
Games are literally the worst benchmark for CPUs because it's realtime.
Cutting an hour off a 20 hour encode job is significant. Reducing the time you spend waiting on compilations is significant. Getting a few more or less fps isn't.

>>61795560
Threadripper is the exact same thing though, it's just an Epyc CPU with two dies disabled

File: [GJM] Urara Meirochou - 08 [65E5361D].mkv_snapshot_14.12_[2017.02.24_11.10.34].jpg (181KB, 1280x720px) Image search: [Google]

181KB, 1280x720px

>>61794312
>a-at least we still got MUH GAYMES
>at 1080p 144hz w/ a 1080 ti
>posting this shit in an HEDT thread in the first place

>>61796218
Threadripper has spacers, they're not actually dead dies. And AMD is charging $1000 not $2000 for it. Also, it has ECC support as does Ryzen.

>>61797195
>Threadripper has spacers, they're not actually dead dies.
That's a complete bullshit excuse though. AMD says the spacers are "needed" to prevent unequal weight distribution. Then why make the CPU so big in the first place. AMD created a solution to a problem they created. It's the same thing with the temperature reading bugs on Ryzen and even previous CPU's. They said it was for fan curve, but they design the fucking motherboard specs. They can tell mobo manufactures what defaults to include in the BIOS. AMD never seems to be completely truthful with their customers, just like Intel.

>>61797273
It's just cheaper to reuse the socket and slot design, stop trying to write a conspiracy theory into everything.

>>61797273
Only some of the Ryzen CPUs have the temp offset (X-series). It's so it can hit XFR clockspeeds.

>>61797273
>Then why make the CPU so big in the first place.
Probably too big to fit on the Ryzen PCB thus too big for the AM4 socket.

Are Threadripper and Epyc socket compatible? If they are then this makes perfect sense.

>>61797564
It's the same socket just with Threadripper half the pins don't do anything. It's a cost saving measure.

File: 20170808_100339.png (627KB, 1080x1643px) Image search: [Google]

627KB, 1080x1643px

>>61789142
Intel BTFO

File: 1445023659746.jpg (35KB, 252x247px) Image search: [Google]

35KB, 252x247px

I'll never understand why /g/ loves arguing over products that nobody on /g/ can afford.

>>61797754
There are several anons on here already planning on buying a Threadraper system.

File: 1500592411099.png (42KB, 150x262px) Image search: [Google]

42KB, 150x262px

>>61795787
>>61796049
Those timings are so fucking loose and still BFTO intel.

>>61797973
Honestly expect the results to be around 3500 at 4 GHz.

>>61797777
>planning
yes , i to have a pcpartpicker threadripper build , unless the cryptos i own go moon or something i am on my more than enough Ryzen 5 1600 anyways

>>61789142
DOA
O
A

File: IMG_4155.jpg (9KB, 233x268px) Image search: [Google]

9KB, 233x268px

Intel should fire their entire marketing team

>>61797754
1800 a Mobo and 16gb 3200mhz is very affordable actually I priced a build at around $900 aud with a noctura cooler to boot

File: dog face.jpg (24KB, 320x292px) Image search: [Google]

24KB, 320x292px

>>61789311

>>61795787
>19 CAS
dafuq

>>61789142
Thank you Intel
Aymmd is finished
Poojets on suicide watch

>>61800366
Poo in joo

>>61789283
Yes because for amdrones the only meanful work is to use cinebench all day.
Toy rupee has already marginal gains on other loads over 7900. When Intel behemots bench are released then you will realize how poor it is like, same as comparing poozen to a real Intel CPU.

File: samus wat.jpg (33KB, 320x332px) Image search: [Google]

33KB, 320x332px

>>61795787
>Windows 8
>128GB of RAM
>CAS19

>>61800416
Cinebench is only ~10% faster for AMD compared to other workloads. It'll still win most of the time.

Kill yourself, BR glueeater.

>>61800442
7960 will probably be double the performance of chink ripper

File: my sides.png (173KB, 640x400px) Image search: [Google]

173KB, 640x400px

>>61800486
In what workload? SYSMark?

Threadripper will be even hotter as it has lower IPC
Intel is more expensive but still way ahead

File: 1501442763271.jpg (14KB, 225x225px) Image search: [Google]

14KB, 225x225px

>>61800486
>>61800586
>not-bingbus with cucked L3
>housefire glue likely not allowing OC above 4GHz without LN2
>beating anything

>>61800586
i9 can't even keep itself cool due to Intel deciding to use toothpaste for thermal interface material.

Threadripper is soldered and has seperate dies and a huge heatspreader. It has no problems keeping itself cool.

http://www.tomshardware.com/reviews/intel-core-i9-7900x-skylake-x,5092-11.html

http://www.tomshardware.com/reviews/-intel-skylake-x-overclocking-thermal-issues,5117-4.html

>>61800417
CB15 reports w10 as w8 , its a bug
128gb of ram becauze muh workstation
also probs not the best ram , explaining the loose timings (likely 2666 ocd to 3200 on the quick)

>>61800586
Meanwhile in reality, the i9-7960X is hotter and uses more power while being weaker and twice as expensive.
As an AMD processor.

300 points more then threadripper, while costing twice as much and probably also twice as hot

>>61797273
>Then why make the CPU so big in the first place

Because it's cheaper to manufacture it the same way as Epyc, and it gives them plenty of room for future expansion. Do you really think that the 1950X is going to be the ultimate CPU for that socket? There'll be a 32-core chip along some day, be it on this process or post-7nm refresh. If they'd chosen to create a smaller package, they'd be limited in how far they could scale up in the future, since Ryzen's basic design of CCXs linked by IF isn't going to change.

>>61795560
>Not even workstation-types need that many threads on a single socket.

If I had the money to build the computer I want, I would likely get a 32 core epyc possibly dual socket and use it as a work station.

But that's just me, I am a hobbyist who renders shit, if epyc works the same as ryzen 8 core, then I can disable cores to keep the heat down and clock 8 of the cores up to 3.6~ and do the work, save reset to full cpu and render. Even if I couldn't disable cores this may be a worthwhile trade off as the majority of the time spent doing shit comes from rendering rather than creating the render. fuck, depending on the program I could render off 31-28 cores and still make new shit on 1-4 left over ones.

>>61800348
>>61800417

People say amd needs expensive ram, which is not true.

What makes ram expensive is timings, you can easily get shit 3200+ ram for cheap, and amd will not give one fuck about the rames time sings till you are well the fuck over the speed ram can currently achieve.

meanwhile on intel, to get those high frames and high averages, you need low cas ram, especially in more modern games that have started to become ram bottlenecked. this ram costs an arm and a fucking leg compared to the shit that can run amd good.

>>61802457

>rames time sings till you are well the fuck over the speed ram can currently achieve

Fun fact: as of BIOS 1401 the x370 crosshair hero actually has some profiles (hand tuned sub-timings by The Stilt) buried deep within the options that are intended to be used with a select few really fast ram kits (3466mhz and up).

File: hqdefault (2).jpg (13KB, 480x360px) Image search: [Google]

13KB, 480x360px

>>61792708
Co2 cooling

>>61789192
>As if that matter
yes it does.

File: 0jad6wsmwjdz.png (1MB, 1920x1080px) Image search: [Google]

1MB, 1920x1080px

>>61789142
>Wow how can AMD even compete?
By being able to overclock without causing a nuclear meltdown.

File: 1462179530831.png (70KB, 350x200px) Image search: [Google]

70KB, 350x200px

>>61804513
DELET THIS RIGHT NOW GOY

>>61789142
>Wow how can AMD even compete?
By having the 1950x hit a 4.1ghz all core overclock at 1.408v and post a Cinebench R15 score of 3,433 :^)

File: 1474400885809.gif (338KB, 538x572px) Image search: [Google]

338KB, 538x572px

>>61804513
>mfw intel release a new SUPER XTREME line with solder being a marked selling point

>>61804513
Thats a big total score

>>61791878
8 cores 16 threads for 250 bucks man, are you in the same year as everyone else?

>>61804513
DELET

>>61804513
muh dick

>>61789142

>inb4 on LN2

>>61789240
Regular consumers would never give a shit about a $2000 CPU, most won't pay half that for a full computer.

File: QNAJc.gif (62KB, 320x240px) Image search: [Google]

62KB, 320x240px

>>61804513

holy shit

>>61797754
The cheapest TR is $550 and you get all 64 PCIe lanes with that. Add in another $4-500 for the motherboard. It's pricey, yes, but definitely something I could afford.
I'd rather go with a 1700 and a new monitor instead, though.

>>61800910
Linux > Windows 8.1 > Windows 7 > Windows 10

File: Intel-Haswell-Refresh-Devils-Canyon.jpg (112KB, 1062x568px) Image search: [Google]

112KB, 1062x568px

>>61804934
Wouldn't be the first time they used unfucking their TIM situation as a marketing point

>>61804513
Why is your ram so slow?

>>61806767
>implying that windows 8 is better than 7 or 10
BWAHAHAH
also show me linuxes version of cinebench

>>61804513
An i9 will still win with oc and better cooling. In fact skylake has a whole lot more overclocking headroom than the competition.

>>61809049
i9 can barely keep itself cool under load at stock. You'd have to delid for sure. And that's pretty fucking dangerous when you're dealing with a $1700-$2000 CPU. Skylake-X is even more difficult to delid than other Intel CPUs.

>>61791175

2146MHz memory?

>>61795787

>128GiB @ 3200MHz
>no

>>61789142
You know what's funny, Amd/Intel and Nvidia are going to be in cahoots all this show of posturing on the world stage every few months or year is just the dance.

Grab your bargain product for price to performance. Whether it be either company, at the time you need to do so and get on with enjoying your PC

>>61808162
because you are a retard

>>61794580
>>61794492
I'm curious, too.

>>61789142
By scoring 4k in cinebench

>>61797754
I have the core i7 6950x.

File: Screenshot_2017-08-08-22-33-01.png (998KB, 1440x1499px) Image search: [Google]

998KB, 1440x1499px

>>61789142
http://wccftech.com/amd-ryzen-threadripper-1950x-benchmarks-leaked-42-faster-competition/
I have no idea op it sure is a mystery

>still arguing which processor brand is better when processor speed hit a huge fucking wall years ago and all the "benchmarks" and dick waving is just a bunch of fucking lies and they goddamn know it, but still want to sell you overpriced shit that isn't any better than their old shit
WHY DO PEOPLE FALL FOR THIS GARBAGE

>>61805073
>250 bucks
where?

>>61794312
Only gamer man babies care about anything you said. Adults are happy to get their video encoded faster and not have to wait for multiple large spreadsheets to open.

>>61797273
>then why make the die so big in the first place

Yeah who doesn't love buying a different socket mobile every time they upgrade their chip?

Why not just buy a pair of slightly older Xeons for the same price?

>>61809911
all the joys of being a neophile drone
new is good! old is bad! consume consume consume!

>>61809931
Probably microcenter. I got my 1700 for $252 after an iobatta jet cashback rebate + some promo codes, though.

>>61810323
Power consumption and upgrade path.

>>61809911
Only Intel hit the wall due to thermals and ringbus design.

AMD have 7nm and infinity fabric they are fine

>>61810323
Worse in every way that matters, that only works for a cheap NAS or something

File: chrome_2017-08-09_00-07-50.png (157KB, 1176x934px) Image search: [Google]

157KB, 1176x934px

>>61800417
>>61800348
>>61797973
It's running 16 ranks of ram, it's amazing it even works at that speed. Ryzen also employs an extremely effective memory interleaving system. So effective, in fact, that 4 ranks of 2667 16-16-16-36-1T can in some cases match, or come close to the performance of 2 ranks of 3200 14-14-14-34-1T. And there is a huge difference in true latency between those two.

https://www.golem.de/news/ram-overclocking-getestet-ryzen-profitiert-von-ddr4-3200-und-dual-rank-1704-127262.html

As I'll mention below, it's also likely using very tight sub timings.

>>61802457
This is absolutely false. The reason ryzen appeared to not care about cas latency and other primary timings at launch is because the default sub timings for their memory controller are insanely tight. The Ryzen IMC natively runs in 1T, which a lot of ram can't even handle. But as soon as you look at the advanced ram sub timings opened up in AGESA 1.0.0.6 it becomes very apparent why you required the absolute best DDR4 IC in production to get up to speed, especially prior to the update when you had zero access to those timings and everything was forced to use Ryzen's insane timings. Simply put, Hynix and Micron memory could not handle the demand put on them.

This being said, Ryzen is extremely memory dependant, and there's a few reasons why.

1. Latency affects how quickly scheduler errors can be corrected by bouncing information from one CCX to the other through the ram.

2. The Infinity Fabric frequency is directly tied to the memory frequency. By overclocking your ram, you're overclocking the entire uncore of your CPU. This is exactly the same as the all important HT bus overclocking back on Phenoms.

Ryzen most likely can't even handle much more than 3600, and even on intel's side, that's highly motherboard dependent getting beyond 3466.

>>61810857
To add to this, Samsung B-Die wasn't that much more expensive than hynix / micron memory until it was actively being sought out by everyone, and not just competitive over clockers. As soon as people started collectively finding what SKUs used what IC's, and actually publicaly announcing it prices were adjusted accordingly.

>>61797754
I do my upgrading with money made flipping cheaply acquired electronics. Like an ECM-55B microphone I just got for $20 and can reasonably sell for $100-150. If I really wanted to, I could get a 1900x + mobo paying only a fraction of the price out of pocket. But that would be entirely dependent on whether or not I could get my hands on a second kit of HOF memory. I would also rather just do other upgrades like a vega + custom water loop.Vega will be 50%+ paid for just by selling the x79 motherboard + xeon I replaced with my 1700.

File: 7700k is so cheap.png (357KB, 820x1661px) Image search: [Google]

357KB, 820x1661px

At least all of the prices have dropped some

>>61796214
I'm giving you a (You) because shills never reply to anything that blows them the fuck out.

Gaymers are cancer,

File: EEFYST000100_1_640.gif (1MB, 640x640px) Image search: [Google]

1MB, 640x640px

>>61800059
>Intlel inside

File: 1489122339460.png (495KB, 1070x601px) Image search: [Google]

495KB, 1070x601px

>>61804513
ONE NOTHING WRONG WITH ME

>>61810857
*segfaults*

File: Screenshot_20170808-232405.png (615KB, 1440x2560px) Image search: [Google]

615KB, 1440x2560px

>>61811374
>Meanwhile in Canada

>>61811381
>hurr durr fuck gamers am I right am I fitting in yet fellow /prog/fugees haHA
You realize without the gaming market desktop PCs are dead, right? Look at how much non-gaming components cost. Quadros are a good example. If there's no volume sale from the gamer market there's no economy of scale and the price goes up ten fold.

If you hate gamers so much, enjoy your Xeon prices on everything.

>>61810857
Thanks for the info anon. We need more guys like you.
t. 1600 owner that got corsair LPX Memgence RAM...

>>61811454
Not only that but the gamer desktop market has been growing for years and it's massive now, and only getting bigger.

>>61795024
That would make them overheat even faster. Contrary to popular belief, in the vacuum of space you cool down very slowly.

>>61811454
>>61811463
I only hate gamers for being retarded when it comes to benchmarks. I play games myself so I am may seem a hypocrite. I just don't throw around screenshots of fps counts to justify one CPU over another like Intel shills do.

>>61811454
Quadros, Xeons, Opterons, these are commercial components. Like any inudstry, commercial components are always a lot more expensive than consumer equivalents.

I'm not saying you're wrong. I'm agreeing with you, but I don't think everything would experience drastic "ten fold" price increases.

>>61811374
>1700x is cheaper than 7700k
I'd honestly go with the 1700x just for the simple fact that it isn't a stuttery housefire.

File: ghelpasd.png (87KB, 256x256px) Image search: [Google]

87KB, 256x256px

>The company's most powerful CPU ever was built for superfans.

Yeah, you're going to need some damn effective cooling with that.

>>61811418
>Everyone compiles
>Compiling is the only reason to have such a cpu.
I'll concede that it is a very serious issue, and AMD / devs need to get it straightened out. But in my digging around looking at different attempted solutions there doesn't seem to be any actual pattern. Some people say disabling the OP cache fixes it for a minimal performance loss, others report that it has no effect. SMT off + one other setting disabled works for a big performance hit, then other people say it does nothing. Some BSD devs found a workaround by relocating the cause of their segfaults to a lower memory layer. Then another group of BSD devs operating the same function in an even lower memory layer by default receive segfaults.

One thing I find bothersome while looking around is no-one is reporting anything about specs / settings / temps / etc. I know phoronix can replicate it at will but shoving 16 concurrent operations down an R7's throat at stock clocks, but I seen one individual (outlier perhaps) that reported a delay in the segfaults when starting his compilation with the cpu cooled down. Ryzen, like all finfet based chips (excluding maybe Intel?) is very temperature sensitive. As far as I've seen, the possibility that the CPUs may be running just a hair too warm has not been looked into. But if it was really something that simple, I can't imagine AMD wouldn't have come out with that as even a possible solution.

>>61811454
It's an annoying bit of posturing that takes place here. But they also have good reason for it when /v/edditors come in and start shitting all over the place. Derailing threads with completely unrelated bullshit.

>>61811459
I mostly blame MSI and their bullshit QVL for that. Prior to learning about the necessity of b-die I had recommended LPX 3200 to a lot of people based on a misleading QVL MSI released. With the latest agesa update, and some fiddling around with sub timings, you should at the very least be able to hit 2933, though.

>>61811435
>same price after conversion
It's not more expensive. You just earn less

>>61811814
They actually earn more before the government takes half of it for socialism.

>>61811794
Will try fiddling with my RAM on a day off. Thanks for the info.

>>61811827
And then they pay less for healthcare, so it evens out.
Not to mention better education, better roads, safer places to live, and a higher quality of life overall for less monye.

>>61794312
>the 7700K is $100 cheaper than the 1800x and runs every game better at stock, much more at OC
The 1600 is $100 cheaper than the 7700k, performs at most 10% worse in games, better at everything else and runs 30C cooler.

File: CINEBENCH Windows 64 Bit_2017-08-09_02-31-39.png (79KB, 1232x870px) Image search: [Google]

79KB, 1232x870px

>>61811852
Couple key things that enabled me to hit 3500 on my ram, twrwrSC, and trdrdSC changed from 1 to 2, command rate 2T, There were a couple things that defaulted to 4 that I loosened to 6, but can't recall them by name. For performance make sure gear down mode, and power down mode are disabled. And set tRRDL to 9, and tRRDS to 7. Make sure the SOC (NB) voltage is set to 1.1v. You may also have to up the ram voltage to help it along. DDR4 is safe up to 1.5v, but I'd stay below 1.4. You could fiddle with procODT settings (do not exceed 60 ohms) but that really shouldn't be necessary for the target speed.

If you have a bclk generator, setting to 105 and using a lower memory multiplier can also be beneficial. It'll also improve core performance overall. I can boot at 3600, but it's not stable. Maybe one day I'll get there. Perhaps after I replace this POS h115i with some better cooling. Also, when stress testing, use HWinfo and check the tctl/tdie temp in the sensors monitor. You don't want tdie to exceed 70c, at 73c you'll hit thermal shut down.

>>61811857
You should move away from the coasts.

>>61811964
Thanks for the info anon :)

cant wait for intel to release that 10 core at 2000 bucks so that someone can get an EPYC 7551P on the same money that has 32 cores

i just wonder how intel is going to survive

File: 4qsfhkk.png (129KB, 1562x332px) Image search: [Google]

129KB, 1562x332px

>>61812026
Yeah no problem.

>>61812072
I just can't wait to see zen 2 launch on that magical 7nm LP process which has a 5ghz target frequency as opposed to the 14nm LPP's 3ghz target frequency. Gonna be sweet. Maybe Zen 3 on IBM wizardry pure silicon 5nm? Good times ahead.

>>61789192
Threadripper has already beat this score. Nice try shill, stay mad

File: 1494824098126.jpg (44KB, 306x548px) Image search: [Google]

44KB, 306x548px

Lol dude intel didn't just get caught with their pants down. They were jacking off to gay pron with a dragon dildo up their asshole when AMD kicked the door down.

File: 1500064796108.jpg (26KB, 377x306px) Image search: [Google]

26KB, 377x306px

>>61789142
>t. marketing division supervisor
Fucks sake. Get some cells between ya damn ears.

>>61811857
You haven't actually ever lived in Canada, have you? Our social experiment is a fucking shitshow.

File: 1494526840474.jpg (137KB, 500x500px) Image search: [Google]

137KB, 500x500px

Wow, they made the cinebench score match the power draw

>>61812905
One (1) performance per watt (what*)

* means ?

>>61802778
Got my 3200 running at The Stilt's fast 3333 preset on bios 9920, feels gud.

>>61812785
By all those metrics, even canada's shitshow is still better than the US.

>>61809911
Amazing how as soon as AMD is driving some progress and competition again the neo-luddides come crawling out of the woodwork to announce Core 2 Duo should be enough for everyone.
>>61811418
http://www.phoronix.com/scan.php?page=news_item&px=Ryzen-Segv-Response

Now fuck off already.

>>61811454
The issue occurs with retarded children/manchildren endlessly screeching that anything that isn't the best for MUH VIDYA is trash, garbage, shit, etc. even if it's better at other tasks.

>>61812785
Living up north? Everything's honestly pretty nice down here in the GTA.

>>61811857
>better education, better roads, safer places to live, and a higher quality of life overall for less monye.
Only one of these is true

Gonna be buying this cpu soon and making a build a build around it since I found out I have 3 bitcoins. Aio cooling or custom loop? Custom loop seems very complicated and I hear it has insane heat problems

>>61816580
Seriously don't waste your money on the i9-7900X. Either get $1000 Threadripper or wait and see if Intel manages to solve their problems with the $1700/$2000 16/18 cores.

>>61816633
This would be my first time going amd in years. The problem with going amd now is that I have to run OSX in VMware at times and I was told amd processors cannot virtualize OSX and that I "need" Intel for that.

>>61789271
yup in few more years, it might actually be wroth it to replace my old trusty i5.

>>61789283
1950x was already oced at 4.1ghz at 1.4v and 4ghz at 1.25v. Golden chips should oc to 4.2ghz, maybe 4.3ghz, but i doubt it. Its obvious that TR already got some optimization over ryzen. NExt Mainstream ryzen refresh will probably bring 4.4-4.6ghz.

File: ryzen_osx.jpg (240KB, 1280x721px) Image search: [Google]

240KB, 1280x721px

>>61816663
looks like people have got hackintosh working and from quick searching vm may not be too far off

>>61789177
BTFO

File: 1448371465623.webm (2MB, 1280x720px) Image search: [Google]

2MB, 1280x720px

>>61789177
Since when are rekt threads allowed on /g/?

(also, does anyone have the longer version of this webm?)

>>61811454
>am I fitting in yet fellow /prog/fugees
What did he mean by this?

>>61817896
'ere you go m8

>>/wsg/1821910