INTEL SURRENDER THREAD

Fuck off. These bait threads are shit when Intel shills make them, and they are still shit when AMD shills make them.

lol intel is two process nodes behind AMD right now, they don't have infinity fabric, all they have is vega theft and lies bout their TDP. AMD says their chips do 95w and they do.... 95. Intel says 65w and they do up to 180.

Intel is being cucked hard and fast.

I used to work for them too, and I like them. But they are 2nd place now.

wrap it up intelailures

Attached: 1537825377281.jpg (812x672, 63K)

>I think we'll be fine m8
sure. your example shows that intel has a huge stock of retarded fanboys who are willing to pay extra money for the inferior performance and the lack of security

Dear Grampa, these threads are created by ARM shills, enjoy your x86 vulnerabilities.

>I don't need 4000 slow ass cores
I guess you don't use any GPU at all, then.

>they don't have infinity fabric
They do have UPI. I don't think there's any fundamental, practical difference.

G-guys.. my 400$ Intel CPU with Spectre mitigations is slower than AMD Phenom.. Remember, there are even more mitigations incoming (eg. 7 new attack variants were just now released)..

Attached: m.png (1076x945, 44K)

never heard of these benchmarks and AMD is effected by the new meltdown and spectre exploits too you fucking amdiscountdicksucker

>new meltdown
There is no such thing as any "new meltdown" vulnerabilities, and AMD was never affected by them.

I feel like switching to Ryzen. It's only affected by Spectre v1 which has the least performance expensive fix. My Intel CPU with all the current mitigations applied has already lost 15-30% performance and it's only the beginning..

I don't really understand these round-about fixes to HT/SMT vulnerabilities. It seems to me the final solution to SMT being fundamentally broken is just to only run threads from the same process (or possibly user) on sibling threads.

You mean, affected by one out of the 25+ discovered so far? While Intel is vulnerable to all the others?
wow AMD wtf

Cringe

INTLEL GPU SUPERPOWER BY 2020

NOOOOAARRRGGGGGHHHHHHHHHHH

Attached: 1542027477262.png (807x745, 205K)

I'm completely willing to accept that this release of the kernel is not optimized. Even if it were, AMD is being really strong and we as end users need it in the market,

based
acidic

based and basedpilled

7nm doesn't matter lol

There's no reasoning with those corporate slaves desu.

>7nm doesnt matter lol
What? what does this even mean? Processing nodes dont matter? So we should go back to 22nm, right? Surely everthing will be just as fast. (it wont)
You dont know what you are talking about. Go back.

intel bros.. I don't feel so good
there won't be many more mitigations coming.. r-right?

Attached: oh no no no no no.png (1817x1325, 130K)

Attached: OY VEY BUY MY CHIPSET.jpg (978x765, 173K)

>intel is two process nodes behind AMD
>I used to work for them too
janitor??

>doesn't want to spend 1k on a cpu
>buys intel
you seem confused

You mean Intel's implementation of SMT being fundamentally broken, right? AMD is unaffected. As usual.

Intel 22nm is better than TSMC 5nm

Attached: 1539398556114.png (500x600, 197K)

800 FUCKING BURGERS FOR A HOUSEFIRE

intel is now at bulldozer ipc

USED

>USED
Yeah, wonder why...

guru3d.com/news-story/intel-cuts-off-diy-desktop-processor-supply-chain-deliveries-for-q4-2018.html

delid dis

delid dis too

Attached: 7s5QTxg.png (800x533, 600K)

Intel CPU prices are going to get retarded in here. As if they weren't bad enough already.

>intel is two process nodes behind AMD
AMD doesn't have a process node. AMD uses TSMC and GF. Intel makes their own chips. Samsung, TSMC, GF and Intel are the companies actually competing to have the best process nodes. Right now Intel and GF are far behind Samsung and TSMC.

Kernel 4.20 has some measures that hit Intel in a really bad way, that's for sure. 20% in many workloads isn't insignificant.

>AMD is effected by the new meltdown
Nope.

These HT problems are probably Intel only. I say probably because the simple fact that nobody's managed to demonstrate that any other CPU architecture has this problem doesn't mean that there could be problems with AMDs SMT implementation.

>this release of the kernel is not optimized.
4.20 isn't released, it's at rc2 nearing rc3. As for "optimized", kernels aren't. There is a Gentoo patch you can apply to any kernels makefile to optimize it for a specific arch. All the distributions kernels and the stock kernel too default to generic x86-64 instructions. You can use the gentoo patch to get march=znver1 for Ryzen or march=skylake-avx512 for Intel if you want to. Doesn't really make that much of a difference, though.

>Phenom
OH NONONO AHAHAHAH

>inteldozer

>shortages up-to mid-2019 at the least
I know this is a bit out there but I really do wonder, given how long 14nm has been around, if Intel's board were looking at the GPU shortages that lasted almost a year thinking "people do seem to buy those regardless of the inflated prices, perhaps we should try to do something similar and triple our profit-margins".

Consider this: NVidia presented earnings on November the 15th and their stock dropped 18.57%. That's a rather ugly drop. One interesting key detail is that NVidia is sitting on millions of 1060/1070/1080 GPU chips. It is interesting how they suddenly found themselves with all that inventory the moment demand from the mining market, which was happy with grossly overpaying, went away..

PRICES AND PERFORMANCE DOESN'T MATTER

Just buy Intel... please

>These HT problems are probably Intel only.
AMD is said to be affected by the latest SMT issue, whatever it was called again, not sure if it was confirmed. But it's not probably reasonable to think of SMT as being inherently insecure to side-channel attacks. When two threads are competing for a large range of computational resources, including cache and branch predictor space, it would almost be strange if it were *not* possible to glean information from it. There's a reason OpenBSD has recommended for years running with SMT off.

>Spectre v1 which has the least performance expensive fix
Spectre v1 has no actual fix, only more-or-less ugly workarounds.

The current 4.20-git linux kernel cross-HyperThread Spectre Variant Two mitigation is specific to Intel CPUs. AMD is not affected by these patches and the resulting slowdown (That doesn't mean there isn't potentially problems with AMD SMT).

>There's a reason OpenBSD has recommended for years running with SMT off.
That's a great point. I have the distinct impression that two of the people working on OpenBSD absolutely do now what the hell they are doing.

Intel are doubling down on keeping their HPC and build partners (desktop and laptop) sweet. Expect Zen 2 to come in like a wrecking ball in 2019. Intel will be fighting tooth and nail with dirty tricks and bribes to keep their HPC and build partners in check.

It was found that out of the 7 SMT hacks recenbtly disclosed only 1 effects AMD in any meaningful way and it is still a non-issue compared to Intel.

Well it won't work, you can't fool giants like Google and Amazon.

Point still standing, though, that adjusting the scheduler to only ever put threads that can ptrace each other on sibling threads would put an end to all such issues, and most likely have insignificant performance problems, since most thread-heavy workloads would fit that model.

You would have said the same thing about giants like Dell and HP a decade ago.

Meant to reply to

Every scummy move Intel tries to do now will easily backfire horribly and will spread all over the internet, see CTS labs and principled shittery.

Private deals between Intel and Google/Amazon won't be blowing up over the Internet without anyone having access to them.

>without anyone having access to them

Attached: 1503807577777.jpg (458x321, 46K)

now post the 5 year

Dammit, I remember AMD being 1 digit per share. People there became millionaires during this hike, huh?

>intelbros
>we
>bros
>we

To make use of the exploits you need hardware access if its an AMD while intel can be done remotely. Patches for AMD are optional they were just released because the paranoic billionaires asked for it.

shut up goy

Attached: 1542271323189.jpg (1800x1800, 190K)