Performance impact of Windows patch and BIOS update (0606) on i7-8700 on ASUS PRIME Z370-A motherboard using Realbench...

>Performance impact of Windows patch and BIOS update (0606) on i7-8700 on ASUS PRIME Z370-A motherboard using Realbench 2.56.
ahahahahaha
>breathes in deeper
AHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHHHH

where were you when intel got btfo and bankrupt?

Other urls found in this thread:

np.reddit.com/r/pcmasterrace/comments/7obokl/performance_impact_of_windows_patch_and_bios/
ebay.com/sch/i.html?_odkw=xeon server&_sop=10&_osacat=0&_from=R40&_trksid=m570.l1313&_nkw=xeon server&_sacat=0
i.imgur.com/Ig2DS9n.jpg
comma.guide/vocative-comma/
reviews.llvm.org/D41723
lists.opensuse.org/opensuse-security-announce/2018-01/msg00004.html
twitter.com/SFWRedditGifs

The interesting question is whether the performance-gimping solution will affect AMD.

>bios update
For what raisan?

mostly for spectre and a few for meltdown.

I just checked for nt msi z370 mobo and there is no update for me yet

It won't, MS disables Kernel VA Shadowing for anything AMD. And Spectre v2 patch requires only lfences for AMD, no perf-damaging microcode updates.

Link?

np.reddit.com/r/pcmasterrace/comments/7obokl/performance_impact_of_windows_patch_and_bios/

he posted some rise of the tomb raider bench mark. those drops in mins. OUCH.

bump

>tfw my Ivy Bridge probably won't get microcode patch through BIOS

bios update makes it slower than software update. makes perfect sense.

ebay.com/sch/i.html?_odkw=xeon server&_sop=10&_osacat=0&_from=R40&_trksid=m570.l1313&_nkw=xeon server&_sacat=0

>all those people trying to get rid of cancer

...

So people speak of Intel vs. AMD, but I'm not up to date: Where does AMD stand performance-wise compared to Intel, maybe even the nerfed Intel with patch.

I would imagine if AMD doesn't employ the aggressive speculative execution that Intel does they would have been a bit slower before the drama hit and now the patches diminish Intel's single-thread advantage?

You can get microcode updates through both.

How do you update your bios?

>tfw 1950x is king of the hill
>AMD relevant again
>DOOM, Quake Champions, new Unreal Tournament is out
>RAM prices through the fucking roof

It's almost like it's 1998 all over again

>mins are lower
>max is higher
>overall scores are virtually identical

wtf is even happening in this chart?

You don't.

Will a Coffee Lake Refresh fix this?

No.

they launched con lake knowing this would happen.

oh fug

Lfences will also hurt performance, but no microcode update is needed. Instead you can set a bit of the MSR to make LFENCES serializing.
To me this kind of indicates that AMD predicted that an attack like this may be found and serialization could be necessary.
Making LFENCES serializing and putting them around critical code WILL HURT PERFORMANCE. We don't know if it will be more or less than the similar mitigations for Intel though.

Also, talk on the LKML suggested that LFENCES are sufficient for variant 1 but retpoline + LFENCES may be necessary to prevent variant 2.

*teleports behind you*
"nothing personell, fuckwit_"

i.imgur.com/Ig2DS9n.jpg

>tfw those threadrippers will go up in price
Even selling my facebook machine wouldn't help me to buy a threadripper.

amazing that the patch manages to improve framerates at the high end. am i being taken on a ruse cruise?

RotB i-is an unreliable b-benchmark!!

still cheaper than $1000 phones

>microcode patch through BIOS
You don't need to.
apt-get install intel-microcode
Done.

this is all margin of error stuff.
as the guy who ran the benchmark reported, the only drop is at the very beginning of the benchmark, where lots of disk access is being done.

max went up, but mins went down. it helped balanced out to keep the average the same... but those drops... ouch did they take a hit.

>53% drop on mountain peak mins
>38% drop on syria
>39% drop on geothermal
>margin of error
>the only drop is at the very beginning of the benchmark, where lots of disk access is being done
i want someone to bench a heavily modded skyrim and fallout 4 with a ton loose files. I'm really to bet big drops.

>he uses apt-get and not the patrician apt
literally JUST

You don't need BIOS updates to update CPU microcode in Windows or Linux. However, my Westmere-EP CPU hasn't had a microcode update since at least 2012 so I am not holding out any hope of getting one now.

funny how not to long ago, like last week actually, people where saying it was a golden benchmark to use. now all of a sudden ITS UNREAALLIBIBLALLLEEE

how long before first CPU models that don't have this fuck up?

retard, the min drop is due to IO, the max "increase" is margin of error.

Since 2017, user

That's besides the point.

fuck off brad

what do you mean? Does it mean Coffee Lake is safe?

>drop is due to IO
oh, look, i/o affecting games when people said it wouldn't affect games.
>hurr only at start
hurrr, not all games are the same. a lot still load shit throughout the game. i really want to see heavy modded fallout 4 and skyrim with a ton of loose files benched.

It's "Fuck off, Brad.".
comma.guide/vocative-comma/

starting to stack up. it is hurting gaming performance.

if i own haswell i cannot expect bios patches right?

reviews.llvm.org/D41723

When manually apply similar transformations to -mretpoline to the
Linux kernel we observed very small performance hits to applications
running typical workloads, and relatively minor hits (approximately 2%)
even for extremely syscall-heavy applications. This is largely due to
the small number of indirect branches that occur in performance
sensitive paths of the kernel.

When using these patches on statically linked applications, especially
C++ applications, you should expect to see a much more dramatic
performance hit. For microbenchmarks that are switch, indirect-, or
virtual-call heavy we have seen overheads ranging from 10% to 50%.

However, real-world workloads exhibit substantially lower performance
impact. Notably, techniques such as PGO and ThinLTO dramatically reduce
the impact of hot indirect calls (by speculatively promoting them to
direct calls) and allow optimized search trees to be used to lower
switches.

intel has claimed they will be releasing microcode updates for all processors since first generation nehalem processors.

Interesting. I'd like to hear about the performance impact of the Intel-specific instructions and the new AMD LFENCE (based on setting the correct MSR bit)

JUST

so intel-microcode will be enough (with kernel patches) on linux? im kind of stupid at this things, i dont even know how to check if my ME is fixed, because the tool crashed for me both on windows and linux.

>no perf-damaging microcode updates
owo what's this?
lists.opensuse.org/opensuse-security-announce/2018-01/msg00004.html

Considering I use my pc as a workstation I have absofuckinglutely no intention of updating.
I'm on windows 7 anyway, if someone wanted to take my passwords they probably already did it.

And to begin with the worst they can do is steal my Steam acc and I'll get that back 5 minutes later from the phone app.

>tfw disabled automatic updates on windows 10 just in time
feels good, man. i'll keep my vulnerabilities.

What's this whole Intel thing about anyway?
I just had it mentioned on Sup Forums and figured you guys could sum it up for a novice.

>4min to 21

>he doesn't know about bitcoin miners

Oh I definitely agree with you here. I'm anxiously waiting for more benchmarks, frankly I don't understand what's taking so long. My point was that this particular benchmark shows that, while there is obviously a performance hit, it is inconsequential in this particular benchmark.

I-i recognize that bulge!

source?

>reddit

no, I ran 3 firestrike benchmarks and all of my scores are slightly higher than before the patch

did you download more ram?

>Windows problems

Kernel compile time was faster for me post PTI.

AMD retards lying their asses off.

no but I will try that and report results

im already at 16gb, you think if i download an extra 8 gb i should be good?

>tripfag shitposting about shills shilling other shills
what a time to be alive

>compiling java
wut

>t. Brainlet

did they already release widespread bios/microcode updates ? only saw the windows and firefox updates

This is a good thing, it will force microshit to get their shit together

haswell has the lowest performance drops anyway

Doom came out in 94, Quake 3 Arena and Unreal Tournament in 99. Athlon64 vs P4HT was on 2004, dont mix things up.

That’s such good news, I’m about to buy an 8700K for a new gentoo box for work and emulation and I was worried! :) thanks for that gentoo friend

>still gonna fall for it
yeh one screencap is all you needed to clear you mind, I guess you should go with it.

You see bigger performance drops on multiple VMs running DBs and handling a high amount network requests. That is where this patch makes a mess. SaaS consumers are pulling their hair out if they had too of a small window in cpu utilization, now they have to upgrade and buy more instances for the same amount of application clients.

Oh I will, thanks bud! Don’t forget that even with the patch applied, 8700K still beats the best ryzen in pretty much every benchmark ;)

Neither Meltdown or Spectre have the potential to install bitcoin miners or do any code execution of any sort.
This is all about being able to read memory it's not supposed to

Why do they update the bios? Shouldn't the windows patch be enough?

so you're a bitter intelet attempting to be cute? ;)

What's hilarious is the moron doesn't even realize this vulnerability pretty much killed shillcoins in one fell swoop if anything.

I didn't want to waste my breath on him though.

No.

One can only hope crypto truly dies with this

4.14.11 currently affects my Ryzen. Apparently 4.14.12 fixes that.

Reality of the matter is nothing can fix this save for making new CPUs.

The patches and updates are just for THESE TWO SPECIFIC attacks. Not for the vulnerability in general. It can't be patched. It's just how CPUs work.

No, microcode updates are needed. The easiest way to get these on newer systems is via BIOS updates.

So if you don't save your passwords on your browser and has JS disabled on any "unsafe" site how can Spectre affect you?

I can only imagine the tears as millions of wallet passwords are stolen.

So my old laptop with the newest patched windows 10 is still vulnerable until lenovo releases a bios update, which they wont.

Same way most other vulnerabilities are exploited; Social engineering. All they need to do is get you to run their code on your computer. Common sense will go a long way to protect you from any threat, including this, but it's not foolproof.

So normies are fucked

Intel microcode will create a method of mitigation. Every single piece of software will need a patch basically. Kernel and hypervisor patches are most important. You can potentially compile software with a compiler that'll automatically use the mitigation.

The JVM and class library are written in C++/Assembly

I think you're wrong
Intel microcode fixes Spectre 2, software patches will fix Spectre 1

I have my browser passwords saved in an encrypted browser file that requires a master password, so I'd say that should be sufficient.

>mins lower
>highs higher
>average almost the same

You're wrong. And you're also wrong to use the word "fix."
Intel microcode updates create new instructions that are necessary to mitigate spectre variants 1 and 2.
On AMD and intel architectures before skylake retpoline patches for the kernel mitigate variant 2.
On AMD LFENCE with a correct MSR bit set will mitigate variant 1 (no microcode necessary)
Intel needs microcode to mitigate variant 1. Intel need microcode on processors after skylake to fully mitigate variant 2. They are adding instructions (discussed in LKML), see the thread linked below

ALL OF THIS REQUIRES SOFTWARE PATCHING, MICROCODE ALONE WILL NOT FIX IT UNLESS YOU DO SOMETHING CRAZY LIKE DISABLE BRANCHING ALTOGETHER

Follow this chain up

It is, though. I remember using it as one of the games I tested my new gtx 1070 with about a month ago. I was trying to OC, and eventually had to ditch it as a benchmark because the min fps was all over the fucking place even at stock clocks. Exact same behaviour as in that pic actually...relatively the same average fps, but wildly different min/max fps.

>So people speak of Intel vs. AMD, but I'm not up to date: Where does AMD stand performance-wise compared to Intel, maybe even the nerfed Intel with patch.
(Pre-patch) IPC is on par(slower in some applications, faster in others) with Intel, but the clockspeeds are capped at about 4GHz max due to the process-imposed limitations(compared to about 5GHz max on Intel CPUs). AMD is generally better if you need high multi-threaded performance because for the price of a 8400(6c/6t) you can get a 1600(unlocked 6c/12t). After the patch intel will still be better in single-threaded workloads except those that suffer worst from the patch.

>Image editing

Geez if only it was the thing I did the most oh wait jesus fuck.

God I want Intel to suck my cock and choke to death on it real hard right now.

If I own Pentium MMX, am I going to receive patch?