What is NEARLY NOT AFFECTED (0-5% slowdown): - Pure computational performance (rendering, number crunching, video games) - Workloads that fit in RAM (small scene rendering, video games, basically everything) - Rare large file operations (>1 GB) So, 95% of computer usage by YOU personally. Your own personal computer is practically not affected. You may never notice a decrease in performance. That's what benchmarks usually measure and they, obviously, show that performance hit is negligible. You are NOT FUCKED.
What is FUCKED UP (5-30% slowdown): - Workloads that don't fit in RAM - Often small file operations (
What will NON-RETARDS do: - Increase chunk sizes to read/write data to disks less often - Increase chunk sizes to move data between instances less often - Cache more, get more RAM so everything caches more by itself - Do less kernel-based IPC i.e through msgsnd, move to shared memory - Do less loopback/socket-based IPC on one machine, it's a lazy degeneracy - MEASURE performance impact on toy cases before spending time on implementation - MEASURE performance impact on real cases before deploying
What should HI-TECH DEGENERATES do: - Install GNU+Linux and ascend - Collect $20 from class action lawsuit - Happily play your stupid games with unnoticeable 0-3% slowdown - Stop creating threads with obvious benchmark results, it only spreads misinformation - Consider masturbating to meaningless number less, get curious, get educated, use this opportunity to learn about computer internals and share the knowledge with concerned non-tech people and your hi-tech fellows
Zachary Long
NOT SO FAST SLOWPOKE
Leo Gonzalez
What will RETARDS do: - switch to AMD
Joseph Martinez
It's the Showdown of the Century!
Lincoln Lee
But don't retards already have AMD?
Jack Diaz
Not yet. they are still on the inferior intel 8xxx series.
Kayden Martin
>what will non-retards do >they'll slow down a shit ton really makes you think
Mason Gray
>linux >games
Blake Morris
>Collect $20 from class action lawsuit Where?
Tyler Bennett
>implying that abolishing lazy computing practices and buying more RAM will slow down anything
Easton Gray
>a degree Literally worthless opinion pajeet
William Kelly
*Although to be fair I agree with you
Dominic Ross
>>parallel computing with a lot of work sharing Does this mean muh xeon phi cluster running stupidly parallel stuff is gonna slow down significantly?
Ryder Green
Check out GNU+Linux game selection on Steam, it'll last you through a few lifetimes. Wine became pretty good, too, I've recently finished Doom on it, ran on native speeds with no bugs and with no tweaking, worked literally out of the box.
Joseph Smith
I play the sims 4 natively on wine fine
Wyatt James
Depends on workload profile. If you spread it across cluster and each node then crunches the data for a whole day - minor slowdown. If the load doesn't fit on one node at all and they have to constantly exchange intermediate results - yeah, you're fucked.
Noah Long
>>Facts from someone with a degree in the field: >source: my dad works at Nintendo
Ethan Stewart
like fine wine xDDDDD
Jaxson Bailey
servers don't need to be patched, server admins have full control over what code gets executed
Adrian Johnson
- I've read on news that there is some mumbojumbo flaw in Intel processors and it's severe, is our infrastructure patched, is our client data safe? - Ackschually, I know better than every security specialist in the world, so no, we won't be patching, I'm in a full control - You're fired, smartass
Logan Wilson
a 5-30% slowdown is unacceptable to a company like facebook. they'll avoid it if they can.
Eli Baker
And if I need to have cores in one socket talking to each other all the time am I totally fucked?
Isaiah Price
So backend servers that only run a certain set of applications could technically skip the patch thus avoiding the performance penalty without any increased security risk, right?
Luke Flores
>without any increased security risk user...
Christopher Gutierrez
from what i've read you need to run malicious code you don't just get hacked out of nowhere
Jace Butler
i play bf2 60 fps on a 4 year old 6300 whos a retard for paying more ;)
Parker Watson
If they communicate via shared memory - no, if you pass messages through kernel - yes, but you can always move to shared memory and implement message passing over it. They'll rather move their databases to AMD than accept that any RCE/minor malware on any machine can lead to a total data compromise without even escalating privileges. Imagine every handgun on a planet turning into a nuclear missile, you can't just ignore that.
Zachary Hall
>They'll rather move their databases to AMD do you even have an inkling of an idea what the cost of that would be
Anthony Ross
Well, the cost of Facebook database breach may cost as much as Facebook itself. They're over-provisioned as fuck, that 0-30% increase in load should be tolerable for them.
Anthony Brown
>a 5-30% slowdown is unacceptable to a company like facebook. they'll avoid it if they can. so facebook will leave user data and internal project X's wide open to theft and destruction which would result in bankruptcy just because they didn't want to buy 30% more computing power? no, they will eat the cost for now, but intel and mexico are going to pay.
William Nelson
tl;dr
Intel is fucked in the server market.
That why the Intel CEO dumped stock.
Justin Sanchez
Sounds like gaming may still be affected by everyone's ping getting worse.
Jayden Morris
>wide open to theft and destruction [citation needed]
Joshua Ward
What's wrong with lazy computing?
Jordan Gray
Nothing, until you have to remove all those hacks because otherwise everything will be 30% slower.
Camden Perry
Explain this the OP.
Jason Carter
Could be a number of things. Personally I don't trust the testing method.
>shill shill shill I'm running my own tests now, my very first run was about 20 points higher than the next four, which were all within a few points.
Whether this caused by latent heat resulting in quicker throttling, I don't know, but it basically isn't the point. It gives a better idea of where the processor really is under heavy loads.
Benjamin Jenkins
These are meaningless numbers with no explanation of what is tested.
Angel Perez
Yes? 30% of non-browser desktop games, not counting using WINE. 100% of console emulated games and browser games. Almost all dx9 and lower games will work perfectly on WINE. So, overall you'll have access to over 70% games you'd have on windows, which is still thousands of games.
Lucas Richardson
>browser games woooow
Charles Powell
You never used cinebench?
Cooper Roberts
It should be fine, as it is all about raw computing power.
Austin Price
Tuning can be a bitch in large systems like you find in companies and research institutions. Especially since much of what needs to be tuned is written in house, and not maintained terribly well.
Nathaniel Sanders
Posted my results
Ayden James
>intel loyalty thread the fuck
Jackson Fisher
I'm amazed that you faggots still think this is actually going to change anything here's whats going to happen >Intel patches bug >Companies that use Intel CPU takes 30% hit >Company does absofuckinglutly nothing because they were over-provisioning from the start as would ANY major corporation do >If they threaten to "switch to AMD" give them a more favourable contract and court them back over. It's not like I enjoy this, but every fucking time you people get my hopes up and nothing happens.
Benjamin Martinez
>- Happily play your stupid games with unnoticeable 0-3% slowdown What games?
Robert Price
>more shared memory >When Spectre exists
We need to share LESS memory
Xavier Hill
...
Jason Mitchell
>full-disk encryption who doesn't use disk encryption nowadays?
William Torres
sry brah, already ripped out my intel CPU and mobo and replacing it with threadripper. intel fucked up. they are done. toast. fuck them like the niggers they are.
Caleb Collins
95% of real-world high-performance computer usage, meaning datacenters have to invest 5-30% more money in ccpus. wowee.
If you lower performance by a third you need to buy 50% more equipment, not 30% more.
Landon Allen
Just how fucked am I if I was planning on using a Windows virtual machine with GPU passthrough?
Jayden Cooper
>95% of real-world high-performance computer usage, meaning datacenters have to invest 5-30% more money in ccpus. wowee.
time to invest in intel stocks goys
Austin Turner
Very. KPTI does hit VMs.
Gavin Gonzalez
...
Levi Martinez
>investing in Intel while every datacenter is in a middle of jumping into the EPYC bandwagon Go ahead, throw your life away.
Henry Ross
...
Julian Baker
Yeah, chinks have already jumped on EPYC cock. I wonder when will kikegle or kikezon will do that.
Gavin Gomez
...
Jayden Williams
the CEO sold out but stocks are soaring already lol
investors are irrational, the world we live in is irrational user.
remember how the capacitor plague crippled motherboard manufacturers in the mid 2000s? me neither lol.
everyone will keep buying intel chips like nothing happened.
Wyatt Reed
Laptop garbage.
Jose Hall
>everyone will keep buying intel chips like nothing happened. Except they'll need more chips and more power for the same computational tasks. they'll just lay off a few thousands.
Aaron Thompson
Okay but will I still be able to watch 4k porn on my iPad?
Camden Nguyen
>- Workloads that rely on many calls to kernel like full-disk encryption using kernel-based crypto facilities So basically, if you are a gentoo user with full disk encryption, FUCK YOU. Did I sum it up correctly?
Carter Bennett
>just rewrite everything!!! Are you fucking retarded?
Jaxon Hernandez
It's 1 fucking VM guys. The cloud corps are up in arms because they run those at scale. Unless you could barely run it before, especially because you'd be passing the hardware directly, it's not enough to matter.
Ryan Nelson
>Switch to linux faggot >But games >BROWSER GAMES LMAO YOU'RE DUMB >O-okay The linux experience
Connor Johnson
Nah, I understood that he pretended to misunderstand what I meant by 'laziness'. I'm all faggot, actually, all programming paradigms are good. I think, around 30*sqrt(2)=42.42640 Well, not exactly, because here it actually hit mostly companies, normal people can play their games normally. No idea, can anyone try this to run the on iOS? Apple admits that it is possible support.apple.com/en-us/HT208394 Use Ryzen then lmao. Nah, just a few fixes if you know what you're doing. And latest Doom (oldest too), and Talos Principle and tons of games on Steam lol. Browser games you can play everywhere, it's nothing to brag about.
Benjamin Jones
>And latest Doom (oldest too), and Talos Principle and tons of games on Steam lol. Browser games you can play everywhere, it's nothing to brag about. See:
Justin Perez
>Companies ARE FUCKED.
Wrong.
>Apple: “Our testing with public benchmarks has shown that the changes in the December 2017 updates resulted in no measurable reduction in the performance of macOS and iOS as measured by the GeekBench 4 benchmark, or in common Web browsing benchmarks such as Speedometer, JetStream, and ARES-6.”
>Microsoft: “The majority of Azure customers should not see a noticeable performance impact with this update. We’ve worked to optimize the CPU and disk I/O path and are not seeing noticeable performance impact after the fix has been applied.”
>Amazon: “We have not observed meaningful performance impact for the overwhelming majority of EC2 workloads.”
>Google: “On most of our workloads, including our cloud infrastructure, we see negligible impact on performance.”
Henry Ortiz
I do the same on my Haswell Celeron. Who's the retard now? :^^^)
Dominic Miller
What do you think their PR statements are gonna say? "Holy shit virtualization is so much more expensive with all this extra overhead and we'll probably even need more mitigations in the future. Fuck our whole business model sucks now?"
Jacob James
>as measured by the Geekbench 4 benchmark They are not even trying to hide anymore.
John Ramirez
>6300 >Haswell that came out after faildozer K
Lincoln Taylor
See
Benjamin Kelly
Well, it's nice than, which means SOME companies are fucked up to 30% while others are not.
Sebastian Taylor
>bf2 >2005 >4 years old 6300 Now who is he fucking retard
Colton Wilson
See what? I'm not sharpshooting, there's a lot of games on GNU+systemd+Linux.
Gabriel Moore
>not able to play latest battlefront ii or any EA game
Fucking disgusting!
Noah Murphy
I think this shitstorm is actually good as it puts AMD back on the map. And drives competition.
And I don't think companies that run Intel will switch to AMD because that would require to change every fucking thing from the board to the code that is mostly optimized for Intel or AMD.
Companies will take the hit but will probably consider having both architectures for fallback.
Luis Clark
60% of the time, it works every time
Carson Watson
When talking about "effects on performance" you're talking about the patches, not the bug itself, which is a mere security flaw, right? Or are you this retarded?
Liam Myers
What sort of idiot would continue to rely on intel after this travesty, on top of the ME disgrace. Intel are just sloppy.
Logan Peterson
What about those intel multicore compiling performance hits? :^)
James Long
Compilers relying on FS performance instead of caching/buffering everything by themselves.
Ryder Allen
T. Inthell
Levi Butler
More like The Slowdown of the Century
James Nelson
>average programmer >knows what they're doing yeah, sure
Carson Garcia
How about video games with server reliance?
Aaron Morgan
How much is emulation going to be impacted, as in CEMU and Android emulation?
Justin Brooks
>Nah, just a few fixes if you know what you're doing. t. unemployed armchair "programmer"
