Reminder that the Intel compiler cheats against AMD CPUs:
realmofespionage.blogspot.com
agner.org
>the Intel CPU dispatcher does not only check which instruction set is supported by the CPU, it also checks the vendor ID string. If the vendor string says "GenuineIntel" then it uses the optimal code path. If the CPU is not from Intel then, in most cases, it will run the slowest possible version of the code, even if the CPU is fully compatible with a better version.
If you want to try:
>To spoof an AMD CPU on VMWare (Player works) edit the VMX and add the following:
>cpuid.0.ebx = "0110:1000:0111:0100:0111:0101:0100:0001"
>cpuid.0.ecx = "0100:0100:0100:1101:0100:0001:0110:0011"
>cpuid.0.edx = "0110:1001:0111:0100:0110:1110:0110:0101"
>cpuid.1.eax="--:0000:0000:0000:--:1111:0100:0110"
>featureCompat.enable = FALSE
Pic related, results from the StarCraft test
>no bumps
Sup Forums - Technology
BTW, apparently, there are some programs that can patch binaries made by the Intel compiler to avoid this fuckery
What other popular programs are actually doing this? Maybe my shit amd cpu isn't that shitty
That's an example for why I'm always saying that proprietary benchmarks are useless.
The only benchmarks anyone should pay attention to are open source, compiled with an open source compiler.
every program compiled with the Intel compiler... lots of games are.
It's a """"feature"""" of every program compiled with ICC.
That's unfortunately a lot.
Is the cripple AMD function still a thing? I thought they'd have removed it once it was discovered.
It's probably just more subtle now.
Really the game developers are to blame for choosing ICC in the first place.
>2015/01
you bet it
this was discovered more than a decade ago...
How the fuck can you even reliably prove it against a large number of important applications? If I was Intel I'd do the same.
And this is why I don't buy Intel products
so why don't devs just use AMD's compiler?
>this bullshit excuse
AMD contributes to GCC and clang/LLVM.
They should be using that.
ISA extensions are standard in x86... the compiler just needs to detect them. and, in fact, it does... but then it cheats anyway
Cause Intel bribes developers to use their tecnology.
Why doesn't AMD talk about this instead of just a few tinfoil hatters on Sup Forums?
ICC can produce some of the fastest binaries for AMD chips family K16 and below. The genuine intel code path issue isn't relevant anymore, though it does still exist in some legacy software, its even present in one of the older versions of Cinebench. Issues in software now aren't being caused by the compiler, but by software developers themselves. AMD had no answer to intel's high end for so long that developers of a lot of professional software started writing code specifically for certain families of intel arch, and it shows now with the launch of Ryzen.
That will likely change given enough time, but its still a reality right now. Not that operations in AutoCAD are terribly taxing when working with 2D, but my Toughbook is edging out my new Ryzen 1700. Pretty impressive show for a several year old laptop. This is simply the developer writing code specifically exploiting architectural features of one processor line so its slower on another. Thats a problem for AMD if they want to win over non datacenter market share.
Reminder that this same excuse was trotted out for Bulldozer and it didn't help because Bulldozer fucking sucked.
Stop making up excuses. You don't even need to. Ryzen is a very competitive architecture. There's no reason to do this.
AMD doesn't have one.
I don't know about the compiler but Intel was fined for bribing manufacturers to not make AMD laptops
This was a thing before Bulldozer.
>AMD had no answer to intel's high end for so long that developers of a lot of professional software started writing code specifically for certain families of intel arch, and it shows now with the launch of Ryzen.
This.
The only solution is what nVidia, throw money at software developers to optimize for Zen.
The issue is that AMD has no money.
>this FUD again
nobody serious uses ICC to compile their software, they use MSVC or GCC.
I think it's just lazy and clueless developers. Like websites that only lets you use html5 features instead of flash when you visit using a specific browser, or a mobile phone. Instead of testing for the required features and only giving you flash when those features are unsupported by the browser.
did you miss the fact that they literally proved the problem was the compiler and not just the way the program was made?
Learn to read, little kid.
How did they prove it? Work on your reading comprehench
>Intel C++ compiler
Why you would willing use a compiler made by a Chip manufacturer when you k ow this kind of fuckery will probably exist is braindead retarded.
Compilers have developers too. There is a lot of shit that is done out of ignorance, or stuff done simply because they're lazy or get super defensive when someone suggests making changes.
bumpo
>Blizzard LGBT programmers make shit games
>Look Intel jews cheat
But jews always cheat
Why don't you blame Blizzard?
Actually, I've read about this as well.
This is a really dumbed-down version of what really went on here:
Intel rigged it's compiler to disable SSE2 instructions on non-intel chips, even if the SSE2 instrunction set was present. This has been patched for a long time, after intel recieved a hefty fine. (Not hefty enough, if you ask me)