• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

AMD Rolls Out GCC Enablement for "Zen 4" Processors with Zenver4 Target, Enables AVX-512 Instructions

btarunr

Editor & Senior Moderator
Staff member
Joined
Oct 9, 2007
Messages
46,518 (7.66/day)
Location
Hyderabad, India
System Name RBMK-1000
Processor AMD Ryzen 7 5700G
Motherboard ASUS ROG Strix B450-E Gaming
Cooling DeepCool Gammax L240 V2
Memory 2x 8GB G.Skill Sniper X
Video Card(s) Palit GeForce RTX 2080 SUPER GameRock
Storage Western Digital Black NVMe 512GB
Display(s) BenQ 1440p 60 Hz 27-inch
Case Corsair Carbide 100R
Audio Device(s) ASUS SupremeFX S1220A
Power Supply Cooler Master MWE Gold 650W
Mouse ASUS ROG Strix Impact
Keyboard Gamdias Hermes E2
Software Windows 11 Pro
AMD earlier this week released basic enablement for the GNU Compiler Collections (GCC), which extend "Zen 4" microarchitecture awareness. The "basic enablement patch" for the new Zenver4 target is essentially similar to Zenver3, but with added support for the new AVX-512 instructions, namely AVX512F, AVX512DQ, AVX512IFMA, AVX512CD, AVX512BW, AVX512VL, AVX512BF16, AVX512VBMI, AVX512VBMI2, GFNI, AVX512VNNI, AVX512BITALG, and AVX512VPOPCNTDQ. Besides AVX-512, "Zen 4" is largely identical to its predecessor, architecturally, and so the enablement is rather basic. This should come just in time for software vendors to prepare for next-generation EPYC "Genoa" server processors, or even small/medium businesses building servers with Ryzen 7000-series processors.



View at TechPowerUp Main Site | Source
 
Joined
May 7, 2005
Messages
32 (0.00/day)
Location
Budapest, Hungary

Attachments

  • avx512.png
    avx512.png
    34.8 KB · Views: 70
Joined
Mar 21, 2016
Messages
2,209 (0.74/day)
At the same time Intel has added support for Raptor Lake and Meteor Lake, with the latter confirming the continued lack of AVX-512.

It's fine Intel fans will tell you why it's great somehow and herald Intel when it introduces AVX-640 as all you need deca-pumping the data in 64-bits across 10P cores at 500w R&D paid for entirely by tax payers stringy $20 billion dollar handout that simply wasn't enough to keep it's marketing team employed thanks to those cheapskate murica tax payer peons not giving them the $40 trillion they originally requested.
 
Joined
Mar 2, 2021
Messages
39 (0.03/day)
Joined
Jun 10, 2014
Messages
2,907 (0.80/day)
Processor AMD Ryzen 9 5900X ||| Intel Core i7-3930K
Motherboard ASUS ProArt B550-CREATOR ||| Asus P9X79 WS
Cooling Noctua NH-U14S ||| Be Quiet Pure Rock
Memory Crucial 2 x 16 GB 3200 MHz ||| Corsair 8 x 8 GB 1333 MHz
Video Card(s) MSI GTX 1060 3GB ||| MSI GTX 680 4GB
Storage Samsung 970 PRO 512 GB + 1 TB ||| Intel 545s 512 GB + 256 GB
Display(s) Asus ROG Swift PG278QR 27" ||| Eizo EV2416W 24"
Case Fractal Design Define 7 XL x 2
Audio Device(s) Cambridge Audio DacMagic Plus
Power Supply Seasonic Focus PX-850 x 2
Mouse Razer Abyssus
Keyboard CM Storm QuickFire XT
Software Ubuntu
I'm actually surprised this patch wasn't done prior to the launch of Zen 4. Typically similar compilation targets arrive 6+ months ahead of Intel releases, so developers have the option in time for the release.

Hopefully not true for their source code.
Which source code? The enabling of a few flags in GCC? (That shouldn't need any commenting)

So many AVX512 instructions. :)
Many?
You're not used to assembly are you? It's very common to have many variants of each class of instructions, like one for each relevant data type and bit width, signedness, logical negation, types of comparison etc.
Just in the original 8086 assembly there were 32 jump instructions (+ 3 return instructions).

-----

Considering how impressive the gains from AVX-512 were for Zen 4, the gains will soon be even larger as the GNU C library adds more AVX-512 optimizations for common core features used in numerous applications.
 
Top