It's fucking over, Kikeripper is out for blood. https:

>The 12-core/24-thread Ryzen Threadripper 1976X ships with a staggering 3.60 GHz core, and 4.10 GHz boost, XFR, and "only" 125W TDP. In comparison, everything 6-core thru 10-core in Intel's Skylake-X HEDT lineup is 140W. The next 12-core/24-thread Threadripper part is the 1956X with 3.20 GHz core, 3.80 GHz boost, XFR, and 125W TDP. Its smaller sibling, the 1956 (non-X), ships with 3.00 GHz core, 3.70 GHz boost, and no XFR, but the same 125W TDP.

File: amadaexciteULTRA.png (210KB, 422x537px) Image search: [Google]

210KB, 422x537px

>>60690736
This is what Zen was designed to do. Bravo, AMD.

>>60690782
Nah, Zen was designed to run 32 cores at 3.0GHz under 200W

This is teenager-tier while Ryzen desktops are infant-tier

File: 1492962036185.png (1MB, 1002x1480px) Image search: [Google]

1MB, 1002x1480px

>double the cores and memory
>only 50% higher TDP
>same clocks as desktop
>ECC
>probably under $1100

Please kill me it hurts to live

File: 1496140523280.jpg (283KB, 1920x1200px) Image search: [Google]

283KB, 1920x1200px

Get ready for assmad fanboys to start making comparisons between a more expensive i9 under LN2 and threadraper on air. For efficiency comparisons, the power draw of the 12c threadraper will be similar to that of a 7700k oc'd to a little over 5ghz.

AMD is being competitive, I like it. I hope they fuck up, or at least make a dent in Intel's sales.

>meanwhile housefire-X 18/16 cores will be at 2.7GHz base clock

Lmao

File: 1487383813301.jpg (11KB, 480x480px) Image search: [Google]

11KB, 480x480px

tfw got to live to see AMD reign supreme
what a timeline

IT'S OVER KIKES

JEWTEL FINISHED AND BANKRUPT

>>60690844
You mean 1998 to 2006?

That wasn't tha tlong ago

File: amd shotgun.jpg (156KB, 633x758px) Image search: [Google]

156KB, 633x758px

>yfw the 10 core skl-x cpu's are hitting 4.5ghz out of the box, the 18 core like to OC as well

I know you pooinloo's thought you'd win, but nope, as again you lose.

No-ones going to buy pooripper or pooga LUL

File: Screenshot from 31-05-2017 22:07:54.png (369KB, 970x610px) Image search: [Google]

369KB, 970x610px

>those base clocks
>those TDPs

Hue

>>60690881
You know Intel shat the bed when their 3.6 base cherrypicked LCC xeon rebrands needs 50% more power than a paltry consumer Ryzen

Intel seems to have sorely underestimated the voltage binning those 80% yields allow AMD to have.

>>60690881
>those clocks
>that TDP
>those lanes(lmao)
>no ECC

The jews dies with a screech, not with a whimper

File: 1477662755244.jpg (51KB, 640x960px) Image search: [Google]

51KB, 640x960px

>>60690907
The binning makes very little difference at this point. It's mostly about the design of the chip. Intel had to come up with around 30% efficiency over AMD's nominal 4Ghz power consumption going from Broadwell to SKL, and in no way was that ever possible. Gonna be funny to see the cooling solutions on those i9 builds kek.

File: lisa_su_smug.png (123KB, 240x286px) Image search: [Google]

123KB, 240x286px

>>60690862
Yeah, on LN2 because nothing else can cool those jizztim abortions.

>>60690736
that's a bog chip

>>60691022
If a threadripper is two 1800X, which are 95W, there's no way there's no binning involved(and there always is due to nature of silicon) at 155W

There's simply no way there isn't binning or process tweaks involved, that would be suicidal.

>>60691022
Threadripper is B2 silicon, not B1 like Ryzen, it's both voltage binning(which you can see in 1800X vs 1700, the former needs a lot less voltage for similar clocks) and fab improvements.

>>60690736
Those part numbers are horrid, AMD marketing is the best fuckup.

>1998X
Kek

>>60691046
Unsure if TR and Ryzen use the exact same die. Also, at 4Ghz all core they will use just as much power, they have per core voltage and frequency regulation to make sure the chips run at a specified power draw. Binning only plays a small part in the difference between SKLX, R7, and TR in efficiency. There were no improvements to the process since Ryzen was already on LPP per AMD themselves and they likely were in production very close to each other.

>>60691127
They're the exact same die but Ryzen has some disabled parts and likely only runs the datafabric at 1:1 instead of 2:1 clocks like TR

>>60690748
That 12-core might beat Intel's 18-core DOAlake-X. The TR 16-core is going to invalidate 3/4 of Intel's single socket Xeons.

If Threadripper really has the full speed infinity fabric then these CPUs are going to outperform Ryzen 7s in games while absolutely smashing DOAlake-X at any fully multi-threaded workload.

The single socket holocaust has begun.

I for one welcome our threadripper overlords.

>>60690748
>"only" 125W
>"only" 125W
>"only" 125W
>"only" 125W
>"only" 125W


AHHHHHHHHHHHHHHH

>>60691149
Source on them being the same exact die? AFAIK we don't know for sure if they really are.

File: 1493663366636.jpg (48KB, 410x416px) Image search: [Google]

48KB, 410x416px

>>60690736
>Kikeripper

>>60690862
>this much delusion
AMD 16 core has higher base clock than even the Intel 10 core, let alone the 18 core.

Dual core boost numbers are irrelevant because nobody buys that many cores to only use 2 of them.

File: AMD-EPYC-Four-Die-with-Infinity-Fabric.jpg (36KB, 957x539px) Image search: [Google]

36KB, 957x539px

>>60691173
Same die, just with more stuff enabled and newer silicon.

File: 1494056927830.jpg (69KB, 494x604px) Image search: [Google]

69KB, 494x604px

my dick can only get so hard

>>60691173
It's basically >>60691193 but with 2 dies instead of 4

>>60691193
How do you think TR will be arranged ? One die up and one die down or one die to the left and another to the right?

>>60691193
I'm just not 100% it's the exact same silicon because I'm assuming there may be something else on die for TR/Epyc to support socket to socket communication. It's definitely the exact same core configuration, but not sure if TR/Epyc is the same silicon and AFAIK it hasn't been confirmed they were harvesting Ryzen chips for TR/Epyc either. Also if it was newer silicon how could they harvest the dies for it from the initial Ryzen batches?

File: dies.png (6KB, 960x540px) Image search: [Google]

6KB, 960x540px

>>60691223
probably one of those

File: AMD-EPYC-Four-Die-with-Infinity-Fabric.jpg (52KB, 957x539px) Image search: [Google]

52KB, 957x539px

>>60691223
One above the other, like this.

The GMI links are on the upper(wider) side of the silicon.

File: MayoLake2.jpg (231KB, 1000x1000px) Image search: [Google]

231KB, 1000x1000px

>>60690736
DELID THIS

File: this kills the SL-X.png (3MB, 1651x1475px) Image search: [Google]

3MB, 1651x1475px

>>60691173
there's only 2 dies: the upcoming 4c Vega-based APU, and Zeppelin, a 8c SOC with specialized Global Memory Interconnect controllers (like HyperTransport or QPI) that facilitate 2-4 die MCMs.

>>60691230
Everything is made out of Zeppelin dies, Ryzen(1 die), Threadripper(2 dies), Epyc(4 dies)

A zeppelin dies already has built in GMI/xGMI links, which is used for socket to socket and die to die communication.

>155W

AYYMD HOUSEFIRES

>>60691246
>The GMI links are on the upper(wider) side of the silicon.

The GMI PHY locations aren't actually confirmed yet:
>>60691257

Do you think 3 die MCM is possible?

>>60691273
True, but they're certainly on the wide side of the silicon, above or below the CCXes, and not to the sides.

Where is the list of instructions it will support?

(That's all these new processors boil down to)

>>60691275
probably not, most stuff requires power of 2s

>>60691275
it definitely fundamentally is, but there'd have to be a new mobo platform with exactly 6 channels to not be wasteful, and AMD/partners might just prefer to sell 24c EPYCs to keep things simple.

>>60691286
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl nonstop_tsc extd_apicid aperfmperf pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_l2 mwaitx hw_pstate vmmcall fsgsbase bmi1 avx2 smep bmi2 rdseed adx smap clflushopt sha_ni xsaveopt xsavec xgetbv1 xsaves clzero irperf arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic overflow_recov succor smca

>>60691286
>b-but AYMD doesn't have m-muh memeX512

>>60690881
intel uses low base clocks to write on paper a low TDP and power draw.I expect them to reach higher clocks on all cores (4.5GHz is I think only on 2 cores), at the cost of higher power draw and the need of a good cooling solution.

Can't wait for proper tests.

PooInteloos on fucking suicide watch.

File: IMG_0147.jpg (66KB, 720x960px) Image search: [Google]

66KB, 720x960px

Seriously, how does Ryzen scale up so well? Their TDP is fucking amazing.

I wonder if it will scale down and keep performance too (on Raven Ridge), do you think it will beat those 45w quad-core i7's at the same TDP?

>>60691301
Yes, Intel bases their TDP of base clocks but their AVX clocks are even lower than base clocks while in normal x86 workloads their all core turbo is some 200-300MHz over base clocks.
The same goes for AMD as well.

>>60691301
we don't even have the tdps for the 14+ core parts, and the 12 core one is already on 165W

>>60691169
12c/24t is equivalent to three i7.

Do the math.

>>60691314
because it's modular, so you basically can use a single die design for everything but the APUs

>>60691283
they probably are. the stuff in the top left of the annotated die shot
>>60691257
looks a lot like the green boxed areas, which makes it a lot more likely that the 4 regions together are the 4 GMI PHY clusters.

>>60691300
a lot of viral marketing here, >>60691299
is refreshing to see actual stats not just hype

>>60691324
on older leaked slides the 12 core has 165W tdp

>>60691321
AMD doesn't throttle AVX2 clocks, but their FPU pipes size is a lot smaller(although a lot more efficient) than Intel's, which ends up with AMD not being all that much slower in AVX2 but once again trouncing in perf/watt

AVX is mostly dick waving and a kneejerk reaction to GPGPU and Nvidia, its use is severely limited to wide parallel vector loads.

>>60691272
Reminder to report spammers.

https://archive.rebeccablacktech.com/g/search/text/%22AYYMD%20HOUSEFIRES%22/

>>60691345
Because it does, 12 core and up to 18(20 unconfirmed) Intels are all HCC silicon with dual ring buses, everything 10 and lower is LCC silicon with one ring bus, the former is naturally more power hungry.

>>60691370
Lel, that explains why they give no HCC info at all, and are delaying their release.
They need time to build up any kind of stock from rebadged Gold Xeons.

Early Skylake-X diagrams(from Intel) said Skylake-X was only supposed to end up at 10 cores (LCC die limit), we can finally properly test how shit the double ringbus architecture is.

What a fucking mess.

File: 1483523720574.png (418KB, 627x627px) Image search: [Google]

418KB, 627x627px

>>60691370
Overclocking on multi ringbus designs is also severely limited because of timing issues. Don't expect there to be much more than stock speeds coming from the 18c.

How is a threadripper different from eg. a workstation or server CPU with a load of cores? Do CPU manufacturers no sell improved single-core performance because it's harder to advance?

>>60691442
What lunatic would even overclock so much cores besides just for dick waving? You'd fucking blow up your motherboard, you'd also ensure your socket melts through your case.

>>60691444
Yes, it's harder to get more performance out of a single core without it using 3 times more power, and there's no protiable market for such power hungry cores.

>>60691460
>and there's no protiable market for such power hungry cores
Games? What else do people who care about cutting-edge consumer CPUs use them for?

>>60691477
No, because you need a ecosystem to support such high current, you'd be looking at 200W 2 core chips.

Nobody but complete retards would want that, and certainly not OEMs

>>60691444
This was the "W" indicator on past Xeons. Also, TR is on the limits of what's possible for single core from the Zen arch so this wouldn't help much. Other Xeons would typically have the first two cores of the first ring set to the highest speed and shift work onto them according to some documentation, which is very similar to what Intel is saying TURBO 3.0 is supposed to be.

>>60691460
Frequency and power consumption is only close to linear in a specific range for most processors, as you go lower you don't save as much power, and as you go higher you use far more. Zen arch has a very broad near linear frequency/volt range and it shows in terms of efficiency. Can't say the same for SKL.

>>60690736
Juranga and Chizow on suicide watch.

>>60691494
>Nobody but complete retards would want that
Like I said, gamers. Just make PSUs with two power inputs.

>>60691605
Don't forget the $300 motherboards and 800W+ PSUs

More power comes with a lot of drawbacks, as said, no PROFITABLE market, OEMs wouldn't touch these things with a 10 foot pole just like they wouldn't touch a POWER9 consumer chip

>>60691453
You could safely do increased max single core turbo, or slight bumps to all core. With 18 cores, you'd only need a 5.5% increase in all core speed to theoretically give you the performance equivalent of another core.

>>60691619
That's why turbo exists, it works on all cores but obviously another 10% increase in frequency at best.

Now a 18core with 2.7Ghz base will likely end up with a 3000MHz all core turbo, but going for 3400, 3800? No way on that many cores, you'd be looking at over 400 watts straight from the socket.

>pooripper 16 core 3.8ghz max boost and can't OC for shit
>intel 18 core has already been overclocked to 4.8ghz on a watercooler

Lul

>>60691640
>400 watts

We're not talking about the rx580

>>60691645
I'd ask for source, but you're just a dumb blind shill so...

It's almost kinda sad, like a dog without an owner sad how desperate he is.

>>60691663
Check computex coverage for yourself you lazy nigger.

>>60691681
>hurr durr lazy nigger
>b-but muh ln2 overclocking
kys

>>60690736
>Mfw E5-2699v3
Yet I want this. Release already. My credit card is ready.

>>60691703
>water
>ln2

Dumb indian

File: wwe.jpg (111KB, 800x1000px) Image search: [Google]

111KB, 800x1000px

literally :
>GLUE MOAR COREZ :DDDD
just wow

>>60691645
>>60691681
>>60691719
>Source: My Ass

>>60691739
So what's wrong if it works?

>>60691719
>4.5 GHz on the 10 core means 4.8 on the 18 core
>https://www.youtube.com/watch?v=Yl3Bn1JycQw

Intel also has some of the finest engineers that worked on it. We can be sure they have tested both tim and solder and that the end products meet their requirements.

Intel isn't out to get me or anyone else. They are in the money making business. And they're good at it, and have the products to back that up.

>>60691640
Also remember that single threaded speeds can disproportionately increase overall task completion speed depending on how much of it is required to be done within a single thread, as per the basis of Ahmdahl's law.

>>60691706
Why use a single socket instead of two? It would end up being cheaper for the performance. I have a E5-2698 v3 but got only one at an absurd price so I could justify it, but 2699v3 in a single socket would normally be paying more for less performance than say, two E5-2673v3 even if it's parallel or single threaded workloads, not to mention being able to pick less dense memory packages for extra savings.

>>60691645
> 16 core skylake
> overclocking to 4.8Ghz on all cores
top kek, a quad-core Kaby lake at 5.0 is already a firehouse, now imagine 16 cores at 4.8.
Don't be delusional anon, don't believe in everything you read on the internet. 4.8 on that many cores would literaly be a firehouse.

File: 1494767421045.jpg (43KB, 630x630px) Image search: [Google]

43KB, 630x630px

>>60691645
>Being so bootyblasted you resort to openly lying

>>60691739
More like intel has 2 more cores and 1ghz advantage over ampoo but they don't hype it like hairy indians

>>60690881
I5 hedt
Can't make this shit up

>>60691778
Intel has hardware core optimization unlike amd which need nuclear voltages to hit 4.0ghz

>>60691762
>that voltage
>that lameass CPU-z score against a fucking 6950X that's running 30% lower clocks
>no temps

That score is pathetic, is that thing throttling?

>>60691739
It's not even that, it's just glueing two CPUs together

>>60691813
>that voltage
>is that thing throttling

What do you think?

>>60691826
And we know how that's going to work out with amd's poo scalability.

Voice of reason ITT, the low TDP is most likely caused by the enabled integrated voltage regulator, which was disabled for desktop Ryzens.

>>60691149
I thought it was zepellin cores

>>60691762
>turboboost active
>only first 3 cores running at 43 multiplier
it's like they don't even want to try

File: time_to_delid.png (327KB, 1508x720px) Image search: [Google]

327KB, 1508x720px

>>60691782
Enjoy your 2.7ghz base clock

File: Intel-FIVR-Fully-Integrated-Voltage-Regulator.jpg (53KB, 571x392px) Image search: [Google]

53KB, 571x392px

>>60691844
AMD has its own FIVR?

>>60691314

I believe this is basically the only thing they hired Jim Keller for.
Make a CPU with i7-matching IPC but focus on scalability.
He provided and now AMD is set for the next few years with intel playing catch up. What a time to be alive.

>>60691859
Enjoy your 3.5ghz turbo (and not being able to overclock)

>>60691871
Yes, this will likely be enabled for threadripper, epyc and mobile platforms.

And more than likely for the 14nm+ refresh for desktop Zen next year.

>The internal voltage regulation (dLDO)

>Zeppelin is the first design in which AMD has extensively utilized integrated voltage regulators. Unlike the fully integrated voltage regulator (FIVR) used in Haswell and Broadwell CPUs, AMD's regulator implementation isn't based on ultra-high speed switching circuitry. The integrated voltage regulators in Zeppelin are ultra-high efficiency digital low-dropout (dLDO) type of regulators. Most of the different domains (cores, caches, data fabric, etc.) have their own dLDOs and they can all be controlled individually.

>Despite the presence of the dLDOs, the consumers can ignore them completely. This is because in the consumer parts most of the dLDOs (all except some of the minor domains) are permanently placed in a by-pass mode. This means that actual regulators are disabled and all of the voltage regulation takes place on the motherboard, just like on the previous generation CPUs and APUs.

When are enterprises dumping all their v3 and v4 xeons for AMD? I can't wait for 100 usd 2699v3 on eBay.

>>60691851
There's no such thing as Zeppelin cores, Zeppelin is the name of the die, Zen is the name of the core architecture.

>>60691917
>server instructions
What do they need differently?

>>60691410
Know as Skylake x emergency edition
Same as P4 furnace

>>60691897
>implying it wouldn't be piss easy to overclock TR and keep good temps with all that surface area and the fact that it's 2 separate dies

>>60691940
Licensing from Intel who wrote them.

File: 1495812218308.png (944KB, 1228x1502px) Image search: [Google]

944KB, 1228x1502px

>>60691897
LMAO the 16-core does 3.9ghz all core turbo, the 12-core does 4.1ghz all core turbo you illiterate shekelchaser
>>60690736
>>60690748

>>60691958
>overclocking 16 core chips

You people are fucking retarded.

File: IMG_3472.png (1MB, 1440x1042px) Image search: [Google]

1MB, 1440x1042px

>>60691964
That's a piece of shit considering Intel's 18 core OC to 4.8ghz

>>60691963
What fucking instruction set extensions though?

>>60691370
ADD MORE BINGBUS ::DDDDDDDDD

Are people this fucking easily baited?

>Intel
NEEDS MOAR BINGBUS!

File: anime_brian_krzanich_placates_th.png (660KB, 698x840px) Image search: [Google]

660KB, 698x840px

>>60691997
>On LN2

>>60691963
there's no additional sets, though. What are you talking about?
also it's intel licensing x86_64 from AMD

>>60692011
Yes, but he's also samefags a lot so you can't really know who's talking to who.

>>60691899
>dLDO
>dildo

>>60692016
Again, nice try poojet. It's been done on an AIO watercooler

File: massive.png (938KB, 1196x605px) Image search: [Google]

938KB, 1196x605px

>>60691970
It's literally 2x 8 cores chips, so the surface area is basically the same as an 1800X

look at noctuas cooler for naples/threadripper

>>60692026
then prove it

>>60692026
On 2 (two) cores. KEK

>>60692029
I'm not talking about the heat, the heat isn't a issue on AMD's platform.

I'm talking about fucking power consumption and how it'll blow up your fucking motherboard.
Do you think these motherboards are fucking made to withstand 300 amps of current?

>>60691888
> Jim Keller
> Jew Killer
I need a quick rundown

>>60692043
both skylake-x and threadripper motherboards have 2x 8pin cpu power, there is enough power for overclocking

File: aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS9DLzEvNTgzNjMzL29yaWdpbmFsLzQxLTY5NTBYLVBvd2VyLUNvbnN1bXB0aW9uLVRvcnR1cmUucG5n.jpg (32KB, 600x450px) Image search: [Google]

32KB, 600x450px

>hey guys lets overclock a 18 core CPU


Yet again, are you people fucking retarded?

File: maxresdefault.jpg (58KB, 1280x720px) Image search: [Google]

58KB, 1280x720px

>>60692026
>Putting 666W though the socket and Mobo is, like, totally possible with an AIO
LMAO

File: 1496318242029.jpg (200KB, 1919x1080px) Image search: [Google]

200KB, 1919x1080px

>>60690736
I'll wait for the fully unlocked chip

>>60691897
How the hell you overclock a hedt CPU without solder? Just kaby lake is a hell without delid I'm not going to delid a 1000$ cpu

File: IMG_3451.jpg (85KB, 552x661px) Image search: [Google]

85KB, 552x661px

>>60692062
That's ironic because pooriper can't overclock

>>60691997
>18 core OC to 4.8ghz
Rajeet please

>>60692025
glad I am not the only one that thought that

>>60692082
my dad works at nintendo too!

>>60691970
Why? It's not the 1800x is thermally limited in overclocking. TR will have no trouble hitting the same overclocks an 1800x will or better given the huge increase in heat spreader size and mass.

>>60691917
((Server instructions))
Come in pal you are drinking too much Intel Kool aid

>>60691899
>dLDO
heh

>>60691022
>picture
Muh dick!

File: notsoldered.jpg (122KB, 1024x536px) Image search: [Google]

122KB, 1024x536px

>>60692082
Nice try, jizztimfriend :^)

File: Screenshot from 01-06-2017 16:02:50.png (166KB, 617x464px) Image search: [Google]

166KB, 617x464px

>>60692067
And the resulting temps.

Do note this platform actually uses liquid metal solder and not mayonnaise

>>60690736
>quadruple processor size
>our new processor is a revolution!!!
>huge TDP
>2010 clock speeds
>requires special mainboard
>requires special cooler

>xeon still better

>>60692111
Test setup

Pump: Alphacool VPP655 (Undervolted, 50 percent)
CPU Cooler: Lian Li
Reservoir: Phobya
Radiator: Alphacool, 24cm (6cm thickness)
Fan: 2x 12cm Noiseblocker eLoop, PWM controlled (up to 1500 RPM)

Soo uhh.. yeah.

Enjoy your thermonuclear reactor

>>60692067
So, am i safe to assume that 4.3 18core will consume about 400Wt under ln2?

>>60691997
>Source: My Ass

>>60692111
We're not talking amd wraith cooler

>>60690736
This is fake, there won't be 10C or 14C models

>>60692118
>xeon still better
>performs the same even while having 50% more cores

>>60692128
Pretty close, assuming you have enough fucking ln2 to cool the damn thing.

>>60692136
>no benchmarks

Lying indian

File: giphy-1.gif (1MB, 352x235px) Image search: [Google]

1MB, 352x235px

>>60692118
Lol the 16-core throws 3/4 of the single socket Xeon line directly into the dumpster.

>>60691970
Did you even read the comment you just answered to?

>>60692067
>240W from CPU alone

Lmao

>>60692149
Is that a pooripper motherboard?

>>60692145
>literally 2 1800X on the same socket and with the same clockspeeds
>no benchmarks

File: 1488192422134.gif (2MB, 249x260px) Image search: [Google]

2MB, 249x260px

>mfw gaymers fucking talking about overclocking SERVER chips.

gaymers are cancerous cunts

>>60692163
>2 480 in crossfire scale 100%

File: greatworkg.png (2MB, 1248x1654px) Image search: [Google]

2MB, 1248x1654px

>>60692160
Lel

>>60692173
>he doesn't understand zen architecture

>>60692166
Thankfully gaymer niggers don't have money for these HEDT chips so the only thing they can do about it is cry.

>>60692136
>>60692149
so what? threadripper should be four times as fast, otherwise it's a trash processor, after all intel could just build a processor the size of a dinner table and beat it again.

>>60692173
even if it loses 20% on the scaling (which it's not going to, because zen scales almost perfectly all the way to 64 cores) it still performs better than most 1S xeons

I WANT Ryzen, but I don't need it! When's Ryzen 2? I won't be able to resist by then.

>>60692173
Yup, officially brain damaged.

>>60692186
>chip bigger than a wafer
I'd like to see someone actually try this.
Assuming they can get one perfect wafer, I think Intel's first perfect wafer was actually somewhere around Nehalem, so don't count on it.

File: 1392100729337.png (73KB, 744x1340px) Image search: [Google]

73KB, 744x1340px

>>60692186
LMAO yeah, with Intel's comical large monolithic die yields. AMD can go all the way up to 32-cores on one CPU trivially while Intel can suck on 0.09% yields :^)

>>60692215
China tried for some shitty ARM cpu and they had to put on chip networking. Surprise, it was both unstable and inefficient.

>>60692118
>1/2 the price

Hm? What was that

What is Intel gonna do with that 18 core? Put it at 230W TDP and 3.3GHz so it even has a chance? These are Xeon Gold parts and according to their spec sheet, they were supposed to be 2.7GHz base at 160W, which is wholly uncompetitive with a 3.5GHz 16 core.. 12% more cores but 30% lower clocks? Complete nonsense. Skylake-X is barely 1-2% faster than Broadwell-E at the same clocks, and fighting AMD on core count with their ridiculously low voltage chips is plain suicide.
And nothing is stopping AMD from releasing a 24/32 core Epyc for the socket as well since the socket is shared on Epic and Threadripper

>>60692118
>smaller tdp than intels counterparts
>half the price
>a good amount of pci lanes
>extremely good at heat dissipation
>higher clockspeeds than any xeon at similar core counts
>better overall architecture

>>60691645
The newer intel chips will only boost 2 cores at the turboboost frequencies.

File: image1.jpg (193KB, 864x645px) Image search: [Google]

193KB, 864x645px

>>60692311
Don't forget ECC ram support so it can throw most of the single socket Xeon line in the dumpster

>>60692296
>And nothing is stopping AMD from releasing a 24/32 core Epyc for the socket as well since the socket is shared on Epic and Threadripper

Besides time and supply, but HCC Xeon rebadges are not releasing like the LCC Xeon rebadges are.

So they have plenty of time.
I'm only worried that AMD doesn't even consider the low clocked 18 core competition and just sticks to 16 core, but higher numbers look good for mindshare.
And EPYC is probably sold out already.

>>60691169
It's "only" 125W because the closest competition is 140W.

Jesus christ look at the size of that thing

File: Intel-Skylake-E-LGA-3647-Socket.jpg (219KB, 1196x967px) Image search: [Google]

219KB, 1196x967px

>>60692427
It's not like Intel's is much smaller

>>60692427
it's a little misleading since the other chip is a mobile (laptop) chip

>>60692449
>that split in the middle of the socket
Not trying very hard to hide the fact they're just gluing processors together.

This is like Q6600-era all over again

>10-core (3+2+3+2), 12-core (3+3+3+3), 14-core (4+3+4+3), and 16-core (4+4+4+4) models
I never though about THAT. Two 4+2 Zeppelins achieveing symmetry using GMI links.

>>60692449
>6 DIMMs per side

Kek

>>60692475
SKL-EP is a ramlet.

File: 1496245787571.png (353KB, 915x678px) Image search: [Google]

353KB, 915x678px

your daily reminder

File: 1472216461903.png (973KB, 801x1500px) Image search: [Google]

973KB, 801x1500px

>>60690817
Who would've thought that Intel would release another Presshot CPU? It's a circle.

>>60691605
Gamers who are willing to spend a lot of money are a small group that doesn't have very good profit margins either compared to enterprise.

File: 1486870038371.jpg (96KB, 624x485px) Image search: [Google]

96KB, 624x485px

>>60690862

>>60692495
They have no choice. Intel are totally fucked until their new arch is ready in 2021. They have no answer for Zen at 8+ cores, much less the unholy beating Zen 2 at 7nm on a high power process is going to give them.

>>60690881
>no turbo boost
What the fuck?

>>60692540
Delid and overclock it, goy!

In that order.

>>60692540
It's Turbo 3.0 aka XFR.

>>60692459
Saves the effort of putting two CPU's in your server motherboard, I guess

>>60692318
>But muh gaymes!

I can see it already

why is techpowerup posting this shit when it was known there would be nine models weeks ago?

>>60692459
Intel's aren't glued together, but they internally have multiple NUMA nodes, and the proposed 6 channel memory parts will have 3 ring buses.

As for the socket, they're split into two because they're easier to manufacture like that.
Pins break all the time in assembly and it's easier to have only one half break instead of replacing the entire array

>>60692611
>3 ring buses.
They moved to KNL mesh for LCC.

>>60692637
>As with other HCC designs in previous years, while the LCC design is a single ring bus around all the cores, the HCC design would offer a dual ring bus, potentially lopsided, but designed to have an average L3 cache latency with so many cores without being a big racetrack (insert joke about Honda race engines). Despite this, Intel shared a die image of the upcoming HCC implementation, as in this slide:

>It is clear that there are repeated segments: four rows of five, indicating the presence of a dual ring bus arrangement.


http://www.anandtech.com/show/11464/intel-announces-skylakex-bringing-18core-hcc-silicon-to-consumers-for-1999/2


Fucking code formatting.

>>60692497
The motherboards for these larger threadripper cpu's are essentially gaming motherboards. The total cost of CPU and motherboard is peanuts comparable to an a deprecated i7 build with Intel which is inferior.
>>60692535
Exactly. Price is on point. Motherboards for AMD are feature packed and reasonably priced compared to Intel's lineup. Most people still game at 1080p which means whatever minor performance difference there is between a higher clocked quad core and say a Ryzen is pointless.

I can't wait to see if I can build an overkill media server to replace my aging one.

>>60692654
>>60692667
Lame. Maybe XCC will have KNL mesh.

File: 1469213542016.jpg (36KB, 420x460px) Image search: [Google]

36KB, 420x460px

>>60692554
>cpu requires delidding to not overheat and run better

I have a Z97 system with an i5. Should I upgrade to a 4790 or 4790K or just start over and build a Ryzen 7 system? I kinda want dem cores, but 8 threads might also help.

>>60692672
I really doubt there's such a huge architectural difference between the two, different interconnect would require a lot of cache and core rearrangement which doesn't sound like Intel at all.

Also KNL mesh is much lower bandwidth per node since it's used for silvermount cores, I have doubts of it being used in a much higher bandwidth/node setup like this

>>60691163
I really hope you're right, I'm looking for a system platform for muh 4K 144Hz gaming, which means I'll need at least 32 PCIe lanes for 2 graphics cards. Intel castrated Skylel-X so the cheapest CPU with enough lanes is $1000, if AMD gives me a Threadripper at $500-600 with 64 lanes and good gaming performance I'll be all over that shit.

>>60692693
No. All you're getting is hyperthreading and some more cache etc with an i7. I have a z97 system here with a 4690k and a 4790k is first overpriced still used. Two not worth the money for whatever slight bump I/you would potentially get. My 4690k runs cool at 4.6 and might go a touch higher. No point.

If you can't help yourself I would look at new Ryzen build and not Intel. Anything new I buy in current year would be AMD based not Intel. I would hold off on any new Intel until they actually have a new architecture which means everything will need to be replaced (if it's competitive).

>>60690736
60c/120t when?
We are going to the moon, bois!

>>60691773
>they're good at it, and have the products to back that up.
And the deep pockets to bribe their way through when the products don't stack up

>>60692816
In 3-4 years.

Gaymen will still be throwing single core benches at you. You know it to be true.

I can't imagine what will happen once AMD decides to stop fucking around and glues together 4 300mm^2+ chips instead of these tiny zeppelin dies.

File: 1494709839630.jpg (111KB, 1394x1017px) Image search: [Google]

111KB, 1394x1017px

>>60692821
me in the screencap

>>60692841
>literally ~1200mm^2 of 14nm silicon
What's the biggest Intel die?

>>60692829
It's funny because threadrippers single core will be better than intel's 16 and 18 core by a wide margin just because they can maintain higher clocks

>>60692029
>that fuckhuge cooler
fucking motherboards are going to need full-cover backplates just to take the load of those things hanging sideways off them in a tower

>>60692844
fugg

>>60692804
I could get a 4790 for 220 eurobux, that wouldn't be so bad

The theme for this year is everyone gets more PCiE lanes. You get some! You too! You! Second time in recent history that AMD just fucks people's shit up like with on die memory controllers and now a gajillion PCiE lanes.

>>60692864
Close to 700mm^2 I think.

>>60692873
Larger sockets means less pressure per mm, the cooler weight is the same but weight distribution on the mobo is far smaller than LGA2066

Anyhow, these motherboards use a lot more PCB layers than mainstream motherboards anyway, mostly due to power requirements, durability is just a plus

>>60692881
Get a 7700k for €300 and OC it to 5ghz. It will take amd shitskins another 6 years to get to that level of single core performance

File: pHrnT.jpg (184KB, 980x602px) Image search: [Google]

184KB, 980x602px

>>60692867
Cry moar shekelchaser. Threadripper going to have more single-core performance AND more multi-threaded performance :^)

>>60692881
220 eurobux even if it was 1:1 with burger bucks is still pricey imo. The newer memelake and baby lake cpu's aren't far from that. You can likely get that or near that price on a sale or deal if you're patient. Ryzen's will eventually get off the new train and start running more sales too.

File: 1473195046525.png (38KB, 795x615px) Image search: [Google]

38KB, 795x615px

>>60692935
Holy shit that housefire

>>60690736
>155w
i would have expected closer to 120w. Do we have a price yet? do we know if it has a management engine?

>>60692953
It's 64 lanes f.a.m

>>60692965
>i would have expected closer to 120w.
>2x95 == 120W

????

>>60692925
But Intel says to not overclock it.

>>60692925
>7700k into a Z97 board
Anon, I...

>>60692936
Well if I buy new it'll be AMD of course. I've been waiting for ITX boards to appear from reputable brands as well.

>>60692925
You forgot the delidding part.

>>60692995
Asrock just released their ITX, and that Biostar ITX didn't have bad reviews at all

>>60692995
Yea waiting to see what happens with ITX or mATX for Ryzen as well. I wonder if it will even happen until they do the APU's later this year or next.

File: 1489816365095.jpg (86KB, 600x450px) Image search: [Google]

86KB, 600x450px

>>60693007
People seem to forget how much faggotry and kikery is involved with the newer intel meme cpus

File: dude.jpg (250KB, 890x1186px) Image search: [Google]

250KB, 890x1186px

>>60692953
>1955 - 10 cores
>1956 - 12 cores
>1976 - 14 cores (non X version doesn't exist?)
>1977 - 14 cores (1977X is slower than 1976X?)
>1998 - 16 cores

>>60693008
that asrock itx board looks pretty clean

wish it had a second m.2 slot though, since there's only ever the one pcie slot for graphics on those boards

>>60693008
>>60693018
I'm actually super torn about what to do.

upgrade current system to an i7?
>only 4 cores
keep current case, get Ryzen ITX?
>a bit bored by the case by now though
ditch everything, get Ryzen mATX?
>most expensive option, plus old parts might not sell

https://www.youtube.com/watch?v=-jWblNZ3Zsw

>>60693094
asrocks threadripper itx board has to be one of the most idiotic motherboards ever made

>>60690736
How much should i bet on that Jewtel will bribe sellers to stay relevant ?

>>60693131
sauce? didn't think threadripper could fit on itx

>>60690748
>thru
Commit suicide.

>>60693128
Personally I plan on getting a threadripper and a new full atx tower. Load it up with used 2TB drives and some external 2-4 2TB external bays for backups etc. Overkill upgrade for my media server. I run a ton of shit though like nextcloud plex sonarr sab and my own usenet indexer.

How much is it going to cost?

You were born in time to see AMD pwn Intel twice. Too bad in a few years nobody gives a shit about Intel, AMD or x86 as everyone will have moved to ARM

>>60693131
threadripper itx?

i saw their x299 itx board which seems pretty stupid but is impressive at the same time

if you didn't give a fuck about lots of pcie devices it'd be cool to have 10+ cores and quad channel ram in an ncase m1 or the like

but i'm hoping for a more workstation-focused threadraper board instead of all this gaymer shit with the rgb lighting and only 3 or 4 pcie slots

>>60693157
you can google it with ease. it fits but its beyond retarded to do so. you get only 16 pcie lanes and cooling it is still a decent issue.

>>60693207
AMD was owning Intel long before Athlon x64 came about.

>>60693207

That's at least 20 years away.

File: 1fx2km.jpg (76KB, 888x499px) Image search: [Google]

76KB, 888x499px

>>60692953
S I N G L E
I
N
G
L
E

S O C K E T
O
C
K
E
T

S H O A H
H
O
A
H

>>60692816
Later this year or next year, with a dual socket EPYC system.

2 x 32c/64t = 64c/128t

>>60693197
My server gonna be running that Xeon X3450 for a few years more, I hope

>>60693242
>>60693229
ignore me, it was x299 which is equally as retarded. how many pcie slots would you need for work? wouldn't you be limiting yourself with bandwidth past 4?

reminder: the x299 chipset is the same as x370, they only changed the name to prevent confusion because of the different sockets.

File: Jim Keller.png (400KB, 816x1020px) Image search: [Google]

400KB, 816x1020px

>>60690736
That's not even my final form.

>>60693291
Yea the Xeon off lease workstations are good deals. Best when you find one with suitable amounts of memory already. I have a couple of them myself for gaming. Bought a couple 4790 based workstations too last year for around 200 bucks too. Running a gamer house practically with all these damn kids. Like hell if I'm going to buy consoles. Gotta raise them right.

>>60693323
you mean 399

>>60693323
x399 you mean? x299 is intel's one

>>60693367
>>60693368
I wish I coul edit my post...

>>60692067
500W incoming? :)

>>60693401
Get that fire extinguisher ready.
Your mobo is not prepared.

>>60693413
lol i wouldnt be surprised if socket melts or something

>>60693433
>CPU gives off so much heat it melts through the case and through the planet into china

Maybe Intel could use one in a few million years to jumpstart the Earth's core once it cools down

File: IMG_4019.jpg (560KB, 2272x1704px) Image search: [Google]

560KB, 2272x1704px

>>60693336
I bought a server board with the cpu and 12 gigs of memory as a package and put it in a Define R2. Gotta have IPMI and the hardware watchdog. (inb4 backdoors)

But this is a bit off-topic. I'm excited to see what the near future brings for Ryzen, cause I will be upgrading the desktop soon.

>>60693448
I'm pretty sure there was a movie about that. A bad one.

>>60693448
make a movie and call it the 10 cores

>>60693456
what do you do when / if you need to use all 8 hdd and all 3 bays? 11 sata connections how? also that cooler looks like it would be better to use an aio, just for placement.

>>60693468
>>60693477
those weren't nuclear bombs, they were overclocked 10 core cpus from intel

https://www.pcgamesn.com/intel/intel-skylake-x-solder

> 4.3ghz is what they managed. They also mention 1.3v being the limit without "serious cooling", whatever that's supposed to mean. These things will be overheating, over priced and will, according to this Asus source, match broadwell-e in terms of OC due to thermal limits. That's an artificial limitation, not a hard one like broadwell's. And yeah, anything that isn't soldered is overpriced because it isn't worth the shoe polish TIM inside it. This was a shady, low, disgusting move by intel. Massive middle finger to their enthusiast customers who dared ask for a better priced product. We got that, and we also got TIM and less PCI lanes.

File: intel ceo JUST.jpg (119KB, 1000x1000px) Image search: [Google]

119KB, 1000x1000px

>>60693514
>mfw 18 core Intel CPUs OCed to 4.8 GHz saved the world

>>60693500
>11 sata connections how?
I'd need to get an add-in card then I suppose. I don't see it happening soon though.

>also that cooler looks like it would be better to use an aio, just for placement.
No no no, gotta be air cooling for stuff that runs continuously. Water coolers aren't reliable enough.

That would work just fine with a stock cooler as well, but I had that Noctua lying around so why not use it there.

File: 446.jpg (89KB, 950x576px) Image search: [Google]

89KB, 950x576px

>>60691859
>2066 lga
good luck with moba price

>>60693575
>moba price

Inflated by $50 because Intel wants that much more money for their HEDT chipsets compared to their mainstream ones.

>>60691952
Bibeline 8============D

File: Screenshot_20170601-114229.png (1MB, 1440x2560px) Image search: [Google]

1MB, 1440x2560px

>>60693336
Haswell-E Xeon chips are already off lease. 12c for $200. $300 for 14. $500 for 16.

2xxx series can handle up to 4 processors combined. A used supermicro 2p board cost $200 and you can throw 2x 12 cores in there for a total of $600, probably significantly less than what threadripper will cost.

16xx series are overclockable but cost slightly more. 2xxx series are overclockable as well via base clock, so those 1.8ghz ends up being 2.7ghz after turbo and oc.

>>60693571
i thought it was blocking a few ram lanes, but its definitely blocking the pcie lane. i suppose if you dont use it its fine

>>60693598
>you can throw 2x 12 cores that will perform worse than threadripper

>The 18 core CPUs are not scheduled until next year. Won't have them for a while. Either way, unless you're using the rig for rendering or encoding to make a living, no need.

t. Raja@ASUS

https://rog.asus.com/forum/showthread.php?93632-Late-June&s=f08ca0f5d8f6d1883b1530987df74d21&p=653561&viewfull=1#post653561


I'm choking here

>>60693598
Those clocks are disgusting.
An 8 core Broadwell would demolish those 12 cores, much less a 16 core Threadripper at 3.5GHz

>>60693639
Maybe for gaming.

24 2.7ghz haswell cores w/ 48 ht would definitely beat threadripper at any "thread ripping"

>>60693655
its obvious, since those are rebranded xeons intel came up after hearing about TR.

>>60693456
Looks pretty clean and functional imo. I dig it. I used an old Antec 300 case for a workstation I picked up. Tossed in 4 2TB and 2 500GB cuda drives striped instead of an ssd. It never gets shut off and the cuda drives are nearly as fast as an ssd anyways (striped). I put another 4 2TB into an external enclosure and striped them as well. I push backups to that plus extra storage. Plan on adding more external and internal at some point down the road. UHD/4k is going to keep file sizes on the large side if you data hoard like me.

File: toothpaste inside.jpg (31KB, 630x372px) Image search: [Google]

31KB, 630x372px

>>60693528
>They've been seeing a 30°C drop in temperatures by popping the top on the chip and using a liquid metal thermal interface.

The jizz TIM meme is real.

>>60693672
Also use 3 times as much power, need a dual socket motherboard and no warranty from China? needing a fucking bclk OC?

Maybe if all you're doing is NAMD which scales linearly with cores, but most prefer both clockspeeds and cores.

>>60690823
Wow well it seems like either of those outcomes will happen, you dumb shit.

>>60693672
>>60693701
Also, Epyc would blow both of them out of the water if you wanted to go that far.

>>60690804
You're forgetting:

>64 (Sixty-four) PCI-e Gen 3 lanes for ALL Threadripper products.

Literally nothing competing with AMD in machine learning now, just plug it with 4 Volta or 4 Vegas and you have a sentient machine.

>>60693655
Great bean counter company, you fucking kikes.

>>60693598
I haven't looked recently so thanks. Those prices look good. You're correct. The off lease equipment will be a much better deal. That's for least a year or two despite Ryzen/Threadrippers awesomeness. Perhaps it might come sooner who knows.

>>60693670
Clockspeed in my usage scenario if I'm not mistaken is best for audio remuxing. Multi par and encoding are fine or better with as many cores and threads you can throw at them. Especially if you have multiple users. Clocks wouldn't be as a huge of a deal except in some circumstances.

File: prasinos_lagos.jpg (48KB, 720x540px) Image search: [Google]

48KB, 720x540px

>>60692935
>1.76v on Vcore for 4.4GHZ

Skylake-X will melt sockets

>>60692554
>Intel CPUs need circumcision.

>>60693772
It's probably symbolic.

>>60693701
Last time I checked ryzen had worse ipc and pref/watt than haswell, which these chips are based on.

Those chips are also highly binned with only 65w tdp with a turbo of 2.5ghz. two of them still consumes less than 155w amd.

Bclk oc is literally entering 102 in the numpad for a free 200mhz boost.

>>60693761
For VM's and any kind of analysis that has single thread functions clockspeed is prefered over pure core count, so something balanced.

But for encoding you're right, that scales almost linearly so more cores are better.

File: aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS9ZLzgvNjY2MDgwL29yaWdpbmFsL2ltYWdlMDAxLnBuZw==.png (72KB, 1112x833px) Image search: [Google]

72KB, 1112x833px

>>60693789
Anon...

>>60693821
That only shows the power consumption of different products without accounting for their performance.

>>60693801
Yuppers. Still running my one machine on Manjaro + GPU passthrough. Going on near two years straight. I can't imagine running a VM on anything clocked less than 4Ghz nowadays. Wonder if that IOMMU group issue (Ryzen) is fixed yet?

>>60693789
>Last time I checked ryzen had worse ipc and pref/watt than haswell, which these chips are based on.
Really? Because in almost all benchmarks, 1800x matches or beats 6900k with lower power consumption. The performance per watt is most definitely better than Haswell.

>>60693887
https://community.amd.com/community/gaming/blog/2017/05/25/community-update-4-lets-talk-dram

>If you’re the kind of user that just needs (or loves!) virtualization every day, then AGESA 1.0.0.6-based firmware will be a blessing for you thanks to fresh support for PCI Express® Access Control Services (ACS). ACS primarily enables support for manual assignment of PCIe® graphics cards within logical containers called “IOMMU groups.” The hardware resources of an IOMMU group can then be dedicated to a virtual machine.

>>60692135
Didn't know you work at AMD, how's VEGA doing?

>>60693829
2x 18 core haswell xeons get 3400~3800 points on cinebench, how much do you think 2x 12 cores will get? meanwhile one 16 core threadripper will get at least 3000 points if their scaling isn't dogshit

>dirt cheap MCM EPYC chips with shitloads of memory
>dirt cheap in electricity costs because LPP node + efficient architecture
Will I finally get good VPS offers which I can afford with my 3rd world shitty wage?
Thank you in advance AMD, you are my greatest ally

File: cinebench-r15.png (10KB, 800x401px) Image search: [Google]

10KB, 800x401px

>>60693789
>Last time I checked ryzen had worse ipc and pref/watt than haswell
It has IPC on par with Kaby Lake, bro.

http://www.zolkorn.com/en/amd-ryzen-7-1800x-vs-intel-core-i7-7700k-mhz-by-mhz-core-by-core-en/view-all/

Has anyone done any recent CCX 4+0 vs CCX-CCX 2+2 benchmarks? The former should have higher IPC which spells nice for Ryzen 3 and Raven Ridge

>>60693965
just look at 1400 vs 1500X at the same clockspeeds

>>60693977
1400 is 2+2 with slice of the L3 disabled

>>60693935
I'm not dedicated enough to research cinebench points but I'm betting 2 would score better.

Also cinebench reference scores are really fucked and are consistently lower than actual scores for some reason.

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

39KB, 826x775px

>>60693965
Here's what you're looking for, but it's from late April so hardly recent.

>>60693965
You cant run asymmetric setups on Zeppelin.
Only (4+4), (3+3), (2+2), (1+1) if you turn off some cores manually

>>60694046
see
>>60694017

Bad gods! Remember the 6 billion.

File: apprentice and second firing door in foreground.jpg (331KB, 1067x1600px) Image search: [Google]

331KB, 1067x1600px

>>60693655
>MFW Intel launches the 18-core just in time to get utterly dumpstered by 7nm HPP Zen 2

>>60694101
*goys

>>60694105
Fuck off.

>>60693296
>wouldn't you be limiting yourself with bandwidth past 4?
Not hard to have more than 4 devices that don't need all that bandwidth.

Video capture cards usually want an x4 slot as does 10GbE networking and I think a raid controller card wants x4 or x8?

With a GPU or two it wouldn't be hard to run out of slots yet only the GPUs want the full 16 lanes each

File: 1489632065735.gif (301KB, 1024x576px) Image search: [Google]

301KB, 1024x576px

>>60694134

>>60694175
true but thats still only a few pcie slots. i can see it being useful for raid and sata expansion like you said though, possibly good for servers in that case

>>60693093
It's a typo, 1976x is blatantly a 12 core with higher top giving it higher clocks. Remember 1700x and 1800x both have x moniker and 8cores so it's not unique per core count.

I9 Chernobyl X 18 cores @ 2.3 165 W water cooling mandatory

Vs

Threadripper 16 cores that will overclock to run all 16 cores at 4ghz I wonder who is going to win

>>60693655
>Either way, unless you're using the rig for rendering or encoding to make a living, no need.
he's right though

>>60694289
Yes, only use for a Xeon is render and encoding, or is he trying to imply these things are fucking useless for crunching, analysis and VMs?

>>60694289
So he's saying everyone should fork over $1000 for less expensive i7-6950K with jizz TIM under the hood instead? lol

File: Intel_Antitrust_Wide.png (223KB, 760x427px) Image search: [Google]

223KB, 760x427px

>>60691773
I agree, fellow Intel enthusiast.

>>60692493
>see manlet chart
I'm 6"2, so I'm good
>see ramlet chart
I have 8GB, so I'm good
>see corelet chart
2c/2t, kill me

>>60694378
Good thing is an upgrade would be practically anything else besides whatever Celeron tier crap you're running.

>>60693128
Evolv Glass mATX is good
specially if you'll put a lot of cooling stuffs on it

>>60694811
Phenom II X2, I fell for the core unlocking meme and it didn't work.

That is a good way to look at it.

>>60693158
Caught that one too when I read the nigger-tier article

File: HAW HAW HAW.jpg (64KB, 640x640px) Image search: [Google]

64KB, 640x640px

>>60693594
Bingbus :----DDD