There aren't too many who aren't.
To me, this is the craziest thing thats happened this gen. Sony's entire next gen plans were leaked in these files not in December 2019 but well before that. December was when rogame found it, but they were out for months before that.
Komachi leaked Oberon 2.0 ghz clocks in July 2019 which was taken from the same github folder. There were several people who contacted the intern who supposedly left his resume and personal information in the github folder, but it was never taken down. Which seems odd, but it seems he didnt put up that folder in the first place.
There was a reddit leak in January which correctly predicted the Lockhart, Anaconda tflops count and said the PS5 was $399 at 8 tflops which is what Ariel was at the time. 1.8 ghz with 36 CUs = 8 tflops.
Albert Penello was also sure that the PS5 was $399 and 8 tflops going back to december of last year I believe, and MS insiders were a 100% sure that the PS5 would be weaker than the Xbox back in April 2019 when the wired article first appeared. No one knew what the PS5 specs were at the time, so this github folder was making rounds at least before the first Wired Article. We all just assumed that third party devs were leaking PS5 specs to MS insiders and MS execs like Penello, but now we know devkits didnt go out until after the first wired article.
Absolutely mind boggling stuff. hmqqq also said he pulled the github folder which is weird. How can he pull a folder that's not his? It's amazing how Sony's plans were leaked so far ahead of time that MS was able to come up with their two console strategy almost two years ago to beat Sony at both performance and price. Sony is probably super pissed about this. lol
What is the hacker asking for?Just for information, the hacker has said that he/she isnt asking for $100 million dollars:
"also not asking for 100m$. if AMD or anyone who wants to settle it down with me for money contact:"
xxXsoullessXxx/GPU-hardware-source-pt.2
Contribute to xxXsoullessXxx/GPU-hardware-source-pt.2 development by creating an account on GitHub.github.com
Just for information, the hacker has said that he/she isnt asking for $100 million dollars:
"also not asking for 100m$. if AMD or anyone who wants to settle it down with me for money contact:"
xxXsoullessXxx/GPU-hardware-source-pt.2
Contribute to xxXsoullessXxx/GPU-hardware-source-pt.2 development by creating an account on GitHub.github.com
Serious question - what is graphics card "source code"? I don't understand what that could be.
Serious question - what is graphics card "source code"? I don't understand what that could be.
Serious question - what is graphics card "source code"? I don't understand what that could be.
Cool. My mind is being blown right now. I have always thought of PC parts and consoles as just the physical things that programs are run on. Like, I have GFX card "y", it can process game code as speed "x". I'm not quite wrapping my head around it. But that is probably not a discussion for this thread. Any reading suggestions to help elucidate this subject?All their specialization in code. That's what makes Xbox a Xbox. All work from their engineers to optimize their systems to work with GPU pipeline , things you saw such as DirectML, Ray traying etc. Driver code on top of AMD GPU basically. Not fully sure what could be done if it includes documentation along with it it could be used for security breach and obviously if it leaks to web competitors such as ARM, NVIDIA etc may copy some of design decisions related to their ASICs.
We don't know what it contains
Cool. My mind is being blown right now. I have always thought of PC parts and consoles as just the physical things that programs are run on. Like, I have GFX card "y", it can process game code as speed "x". I'm not quite wrapping my head around it. But that is probably not a discussion for this thread. Any reading suggestions to help elucidate this subject?
Hardware is made by writing Verilog/VHDL code these days. Think about it like a text representation of all the wires and gates. The actual layout of the wires in the chip based on this is then later done either by hand or by an automated process.Serious question - what is graphics card "source code"? I don't understand what that could be.
possible a job? its often a good way to get hired
To me, this is the craziest thing thats happened this gen. Sony's entire next gen plans were leaked in these files not in December 2019 but well before that. December was when rogame found it, but they were out for months before that.
Komachi leaked Oberon 2.0 ghz clocks in July 2019 which was taken from the same github folder. There were several people who contacted the intern who supposedly left his resume and personal information in the github folder, but it was never taken down. Which seems odd, but it seems he didnt put up that folder in the first place.
There was a reddit leak in January which correctly predicted the Lockhart, Anaconda tflops count and said the PS5 was $399 at 8 tflops which is what Ariel was at the time. 1.8 ghz with 36 CUs = 8 tflops.
Albert Penello was also sure that the PS5 was $399 and 8 tflops going back to december of last year I believe, and MS insiders were a 100% sure that the PS5 would be weaker than the Xbox back in April 2019 when the wired article first appeared. No one knew what the PS5 specs were at the time, so this github folder was making rounds at least before the first Wired Article. We all just assumed that third party devs were leaking PS5 specs to MS insiders and MS execs like Penello, but now we know devkits didnt go out until after the first wired article.
Absolutely mind boggling stuff. hmqqq also said he pulled the github folder which is weird. How can he pull a folder that's not his? It's amazing how Sony's plans were leaked so far ahead of time that MS was able to come up with their two console strategy almost two years ago to beat Sony at both performance and price. Sony is probably super pissed about this. lol
Is this pure speculation from you guys' part?i really doubt it affected ms plan, the one that changed its plans was sony since they went above the 8tf they always targeted. ms expected 8tf and went with 4 and 12 and sony made adjustments
Isn't this stuff normally patented?All their specialization in code. That's what makes Xbox a Xbox. All work from their engineers to optimize their systems to work with GPU pipeline , things you saw such as DirectML, Ray traying etc. Driver code on top of AMD GPU basically. Not fully sure what could be done if it includes documentation along with it it could be used for security breach and obviously if it leaks to web competitors such as ARM, NVIDIA etc may copy some of design decisions related to their ASICs.
We don't know what it contains
some is true, some of it is speculation. the github leak did have the ps5 gpu test results. and there was definitely a leak in jan 2019 that first mentioned the ps5 being an 8 tflops gpu which lines up with the tests in the github leak's ariel folder.Is this pure speculation from you guys' part?
I mean this is insane if true? Does this strain Sony, MS relation with AMD?
Serious question - what is graphics card "source code"? I don't understand what that could be.
library ieee;
use ieee.std_logic_1164.all;
use ieee.numeric_std.all;
entity half_adder is
port (
i_bit1 : in std_logic;
i_bit2 : in std_logic;
--
o_sum : out std_logic;
o_carry : out std_logic
);
end half_adder;
architecture rtl of half_adder is
begin
o_sum <= i_bit1 xor i_bit2;
o_carry <= i_bit1 and i_bit2;
end rtl;
It's Verilog Hardware Description Language, a syntatical representation of boolean logic that dictates transistor layout in chipsets. It's an actual source code that looks like this:
Code:library ieee; use ieee.std_logic_1164.all; use ieee.numeric_std.all; entity half_adder is port ( i_bit1 : in std_logic; i_bit2 : in std_logic; -- o_sum : out std_logic; o_carry : out std_logic ); end half_adder; architecture rtl of half_adder is begin o_sum <= i_bit1 xor i_bit2; o_carry <= i_bit1 and i_bit2; end rtl;
The above is the VHDL code for a half-adder circuit. The above source code describes the following logic circuit, which performs 2-bit addition:
VHDL are what is used to frabricate chips. FPGA cores, for example, which are processors that can "change shape" into other processors, use VHDL source codes to "change."
Nothing directly, even harder without the documentation. However it can be used to give insight into how modern GPUs are made and the tricks there in. Also AMD specific tricks.If the hacker leaks this information, what can actually be done with it? Please forgive my ignorance
Serious question - what is graphics card "source code"? I don't understand what that could be.
Verilog which is an HDL (Hardware Description Language). Makes making hardware easier, testable, verifiable, etc... Also soft-core which is a digital representation on an emulator all the way down to an FPGA. As for what hackers can do, other than better understanding how it's made and AMD Magic. A very big issue is security vulnerabilities exploits found from this leak.
Yes it could result in jailbreaking the consoles if it's a hardware exploit/non-patchable.What exactly can someone do with the source code? I am asking about random people as I imagine the competition (nVidia) could use it to learn a trick or two from AMD. Could someone potentially use it to 'jailbreak' the console?
Is this pure speculation from you guys' part?
I mean this is insane if true? Does this strain Sony, MS relation with AMD?
Vhdl and verilog are actually different languages.
I'm surprised they use vhdl (even though I've never used it) I always imagined that it was kind of low level like verilog. At the company I work at (one of their competitors) , we use system verilog which is somewhat higher level.
I'm sure they use both, but could VHDL be for control, i.e. Security and Performance?Vhdl and verilog are actually different languages.
I'm surprised they use vhdl (even though I've never used it) I always imagined that it was kind of low level like verilog. At the company I work at (one of their competitors) , we use system verilog which is somewhat higher level.
Ah ok. got it.I didn't mean to imply they weren't. I always confuse the V in VHDL to stand for "verilog" though.
I assume it's the test bench capabilities of VHDL that make it the language of choice here. Over in Sega Genesis homebrew, VHDL reigns supreme because it allows for unit testing while reverse engineering components through the test bench.
Ah ok. got it.
Well, nowadays, the industry moved to something called UVM for testbenches which I think is SystemVerilog only (gosh, don't quote me on this though,) so that doubles my surprise. But change is also pretty hard so maybe they haven't moved everything to that. Or the synthesizeable RTL is in VHDL and only verification is using SystemVerilog.
I'm just guessing since I obviously never looked at this repo, haha...
No implications on your past comments as I've never read them. I was asking the other 2 fellows on their own posts but thanks for your insight.I'm not 100% sure what is being implied here, so forgive if the comment is a non-sequitur.
I think I've posted this before, but in my time with Xbox I was never exposed to a direct leak from either a developer or AMD about the competitions plans. People in a position to know things at this level of detail are all professionals and have business relationships with both companies that have persisted, at this point, for decades. Nobody with influence is interested in being the giver, or receiver, of this type of information. A big part of the reason these consoles are getting more homogeneous is because developers give similar feedback. And when one platform deviates too far, developers tend to push back to get things more inline.
Obviously, both companies talk with Developers and suppliers, understand what's available from a SOC perspective and what are bottlenecks for games. Some things (like SSD's) were basically inevitable for both consoles given that data transfer speeds were gating CPU and GPU even late in the current gen. There is nothing that required "leaks" to assume SSD would be a big player (you probably saw tweets from several prominent Xbox employees talking about SSD testing back in like 2014/2015)
Therefore - it's not hard to speculate or have some sort of theory about what the competition might do when you've been doing this for a while. Not due to leaks but simply by understanding what technology is available and at what price points. When you go through a process to decide your plan, you also put together game theory on what the competition might do based on their history, what they might see as success factors, and the company's strengths/weaknesses. Anyone with a strategy background understands a SWOT analysis. My personal SWOT for Sony when I was there was 8.4 +/- 10% at $399 and actually went as far back as 2016. It's highly likely that the current team adjusted their plans over time as they understood more about the development of Series X.
I can't of course speculate what Sony's SWOT was for Xbox. I'm sure they assumed based on comments from Phil that performance was a big factor in their console plans, and I have every reason to believe that Sony had the same amount of knowledge about the Xbox plan as Xbox
Point being - the Github stuff was really interesting to the community, and likely frustrating for Sony, but came far too late for either company to affect their plans and hold a 2020 ship.
Again, sorry if this wasn't the point you were making.
At the companies I've worked at, I've seen a mix of VHDL and Verilog. They both have pretty much the same capabilities, including for test benches. Typically it comes down to comfort among the engineers on the project. I've even seen projects that are mixed. Some VHDL modules and some Verilog modules. Even Verilog testbenches for VHDL modules.I didn't mean to imply they weren't. I always confuse the V in VHDL to stand for "verilog" though.
I assume it's the test bench capabilities of VHDL that make it the language of choice here. Over in Sega Genesis homebrew, VHDL reigns supreme because it allows for unit testing while reverse engineering components through the test bench.
Yea, that makes sense. It's kind of interesting that VHDL is more popular among hobbyists since it feels like verilog (and its relatives) are more popular in industry.Well I'm not an EE, I'm entirely self taught on electrical engineering from homebrew and hobbist circles over the years, so I'm definitely not aware of what tools are used currently in chip design. I have some friends who work for intel in this field, but they don't really talk to me about the weeds of their job. I learned VHDL, because that's what everyone around me in the homebrew circles I run in uses, and I didn't want to be left out.
Yea, that makes sense. It's kind of interesting that VHDL is more popular among hobbyists since it feels like verilog (and its relatives) are more popular in industry.
Yes, this is spot on. As I mentioned earlier, it's usually based on team comfort or what the company has standards for. So yea in the type of thing you're talking about it would make sense for the most knowledgeable person/people who get the project off the ground to set the standard.It might honestly be a case of VHDL is what the most knowledgable person in the scene knew at the beginning, and thus it set the trend. Like how in Sonic hacking circles, ASM68k is still widely used (and many of that assembler-specific macros are dished through out the community source codes) owing to that being the assembler everyone used in 1998, not because it's necessarily any better or anything.
No. Nothing is public yet.Holy shit, I missed this thread.
What in the ever-living fuck?
Did we learn how many CUs Big Navi / Navi 21 has?