Wednesday, September 11th 2019
Intel Says Its Upcoming Gen12 GPUs Will Feature Biggest Architecture Change In A Decade
Intel is slowly realizing plans to "one up" its GPU game starting from first 10 nm Ice Lake CPUs that feature Gen11 graphics, equipping users of integrated GPUs with much more performance than they previously got. Fortunately, Intel doesn't plan to stop there. Thanks to the recent pull request found on GitLab Mesa repository, we can now expect to receive biggest GPU performance bump in over a decade with the arrival of Gen12 based GPUs, found on next generation Tiger Lake processors.
In this merge request, Francisco Jerez, member of Intel's open source Linux graphics team, stated the following: "Gen12 is planned to include one of the most in-depth reworks of the Intel EU ISA since the original i965. The encoding of almost every instruction field, hardware opcode and register type needs to be updated in this merge request. But probably the most invasive change is the removal of the register scoreboard logic from the hardware, which means that the EU will no longer guarantee data coherency between register reads and writes, and will require the compiler to synchronize dependent instructions anytime there is a potential data hazard..."Planned for release sometime around 2020/2021 (with Tiger Lake), Gen12 graphics features a complete overhaul of Execution Unit in a way we haven't seen since i965 debut. There will be less hardware logic that checks data for coherency, possibly resulting in lower latency and higher performance. That workload will shift from logic built into hardware, to compilers for them to figure out if data is correct or not, resulting in less wasted GPU clock cycles dedicated to that function.
Source:
Phoronix via HotHardware
In this merge request, Francisco Jerez, member of Intel's open source Linux graphics team, stated the following: "Gen12 is planned to include one of the most in-depth reworks of the Intel EU ISA since the original i965. The encoding of almost every instruction field, hardware opcode and register type needs to be updated in this merge request. But probably the most invasive change is the removal of the register scoreboard logic from the hardware, which means that the EU will no longer guarantee data coherency between register reads and writes, and will require the compiler to synchronize dependent instructions anytime there is a potential data hazard..."Planned for release sometime around 2020/2021 (with Tiger Lake), Gen12 graphics features a complete overhaul of Execution Unit in a way we haven't seen since i965 debut. There will be less hardware logic that checks data for coherency, possibly resulting in lower latency and higher performance. That workload will shift from logic built into hardware, to compilers for them to figure out if data is correct or not, resulting in less wasted GPU clock cycles dedicated to that function.
67 Comments on Intel Says Its Upcoming Gen12 GPUs Will Feature Biggest Architecture Change In A Decade
However, they are fucking liars. My bad, link me to forum rules plz.
ARM - don't even go there, Apple literally smashes everything including Intel in the CPU+GPU space in the sub 15~10W space! That's debatable, though it's hard to find too many holes in that argument.
Why aren't you so eager to wait for Zen2's real competitor from Intel? :-) It should arrive more or less at the same time as Zen2 mobile SoCs.
At this very moment Intel has a more powerful mobile SoC. Why can't you just admit it? Why is it so painful to you? And what is the maximum resolution for A12? How many screens? :-)
You don't get this, right? Somehow you just can't grasp the idea that computers are about functionality first, performance second. For you it's all about benchmarks and smashing. :-)
I don't mind Intel ramping up their IGP performance, perhaps with competition we will see some AMD CPU with HBM on some market (gaming notebook or NUC)
APUs are usable without this and have been for over a decade. You're right, its a res bump, and therefore its about 0% interesting :) Even with its efficiency, its another baby step yawnfest.
Intel is said to be addressing this in Xe drivers. It's going to be a proper, unified gaming driver (from IGP to big cards). I find this quite surprising as well. I don't know why we're not getting small gaming PCs to compete with consoles.
Of course you can get a powerful NUC or something from Zotac. But it's still using Windows or Linux. I'd rather have a tiny OS optimized for games - just like on consoles.
Steam OS is just an average Linux distro with a fancy skin. It's fairly normal today for software developers, analysts or admins to work on 3 screens, so I'm not sure what you're trying to say... Even most accountants I've seen prefer at least 2.
Windows is awful at scaling, which means office screens are mostly ~20-24" 1080p (or a much better 1920x1200). As that gets fixed, 4K will become the new norm.
MacOS scaling works properly and their ecosystem has already moved to (and past) 4K. APUs are way too big, way too power hungry and not as polished as IGP. Intel could make a bigger IGP as well - every CPU could get something from Iris range. But what's the point?
Intel is slowly improving their IGP to address current needs of non-gaming consumers and business.
I'm sure you know the answer to that, care to share what's the maximum resolution (not iPhone screen resolution!) actually supported by the A12 GPU? Never mind that A13 is already in mass production & will probably sell
10x10000x the (ICL) units Intel moves by then, as the new iPhones launch :rolleyes:You're saying that with a straight face, really? I guess the iPhone or iPad is just a toy to you, huh ? Not to mention you claimed that Intel's IGP is @par & equally(?) power efficient as any of it's competitors - both claims are dubious at best :shadedshu:
They have no technology for radical improvement in performance per core.