AVX512-IFMA cpus

[bsquared]

I am trying to figure out exactly which Intel cpus have the AVX512-IFMA instructions, for a new laptop purchase.

I believe the i3/5/7 10xxxG processors have them (e.g., i3-1005G1). And I think the older cannon lake's, e.g., i3-8121U.

On linux, the output of cat /proc/cpuinfo should show avx512ifma somewhere in the flags section. I'm not sure how to check on windows... maybe cpu-z will list the flags.

Anyway, looking for verified cpu model numbers so that I buy the right thing.

[EdH]

I don't know if this is helpful (or you already have it), but the following link will take you to a datasheet on the 10xxGx Processors, which contains the following on page 38:

AVX512IFMA (Integer Fused Multiply-Add): Fused multiply-add of integers using 52-bit precision.
. . .
Note: Intel® AVX-512 may not be available on all SKUs.

Maybe finding the proper SKU is necessary?

Link: https://www.intel.com/content/www/us...eet-vol-1.html

[bsquared]

Originally Posted by EdH

Maybe finding the proper SKU is necessary?

That is what I'm afraid of... I'm hoping someone has a system with IFMA that can let me know exactly what they have.

Or maybe it's possible to specify that instruction set as a filter for amazon EC2 instances or Azure or something? I mostly want it for code development for that instruction.

[kriesel]

https://www.sisoftware.co.uk/2020/09...and-tigerlake/ indicates Ice Lake and Tiger Lake add IFMA-52.
If you want to add detection into an application https://github.com/Mysticial/FeatureDetector may be useful.

I think CPU-Z does not give that level of detail. Following is the processor portion (sans sensors) of a text report from it, for an i7-8750H.

Processors Information
 -------------------------------------------------------------------------

Socket 1            ID = 0
    Number of cores        6 (max 6)
    Number of threads    12 (max 12)
    Manufacturer        GenuineIntel
    Name            Intel Core i7 8750H
    Codename        Coffee Lake
    Specification        Intel(R) Core(TM) i7-8750H CPU @ 2.20GHz
    Package (platform ID)    Socket 1440 FCBGA (0x5)
    CPUID            6.E.A
    Extended CPUID        6.9E
    Core Stepping        U0
    Technology        14 nm
    TDP Limit        45.0 Watts
    Tjmax            100.0 °C
    Core Speed        1396.2 MHz
    Multiplier x Bus Speed    14.0 x 99.7 MHz
    Base frequency (cores)    99.7 MHz
    Base frequency (ext.)    99.7 MHz
    Stock frequency        2200 MHz
    Max frequency        4100 MHz
    Instructions sets    MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, EM64T, VT-x, AES, AVX, AVX2, FMA3
    Microcode Revision    0xD6
    L1 Data cache        6 x 32 KBytes, 8-way set associative, 64-byte line size
    L1 Instruction cache    6 x 32 KBytes, 8-way set associative, 64-byte line size
    L2 cache        6 x 256 KBytes, 4-way set associative, 64-byte line size
    L3 cache        9 MBytes, 12-way set associative, 64-byte line size
    Max CPUID level        00000016h
    Max CPUID ext. level    80000008h
    Cache descriptor    Level 1, D, 32 KB, 2 thread(s)
    Cache descriptor    Level 1, I, 32 KB, 2 thread(s)
    Cache descriptor    Level 2, U, 256 KB, 2 thread(s)
    Cache descriptor    Level 3, U, 9 MB, 16 thread(s)
    FID/VID Control        yes


    Turbo Mode        supported, enabled
    Max non-turbo ratio    22x
    Max turbo ratio        41x
    Max efficiency ratio    8x
    Speedshift        Autonomous
    O/C bins        none
    Ratio 1 core        41x
    Ratio 2 cores        41x
    Ratio 3 cores        40x
    Ratio 4 cores        40x
    Ratio 5 cores        39x
    Ratio 6 cores        39x
    IA Voltage Mode        PCU adaptive
    IA Voltage Offset    0 mV
    GT Voltage Mode        PCU adaptive
    GT Voltage Offset    0 mV
    LLC/Ring Voltage Mode    PCU adaptive
    LLC/Ring Voltage Offset    0 mV
    Agent Voltage Mode    PCU adaptive
    Agent Voltage Offset    0 mV
    TDP Level        45.0 W @ 22x
     TDP Level        35.0 W @ 17x

And for i5-1035G1

Processors Information
-------------------------------------------------------------------------

Socket 1            ID = 0
    Number of cores        4 (max 4)
    Number of threads    8 (max 8)
    Manufacturer        GenuineIntel
    Name            Intel Core i5 1035G1
    Codename        Ice Lake
    Specification        Intel(R) Core(TM) i5-1035G1 CPU @ 1.00GHz
    Package (platform ID)    Socket 1526 FCBGA (0x7)
    CPUID            6.E.5
    Extended CPUID        6.7E
    Core Stepping        
    Technology        10 nm
    TDP Limit        15.0 Watts
    Tjmax            100.0 °C
    Core Speed        1695.9 MHz
    Multiplier x Bus Speed    17.0 x 99.8 MHz
    Base frequency (cores)    99.8 MHz
    Base frequency (ext.)    99.8 MHz
    Stock frequency        1200 MHz
    Max frequency        3600 MHz
    Instructions sets    MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, EM64T, VT-x, AES, AVX, AVX2, AVX512F, FMA3, SHA
    Microcode Revision    0x70
    L1 Data cache        4 x 48 KBytes, 12-way set associative, 64-byte line size
    L1 Instruction cache    4 x 32 KBytes, 8-way set associative, 64-byte line size
    L2 cache        4 x 512 KBytes, 8-way set associative, 64-byte line size
    L3 cache        6 MBytes, 12-way set associative, 64-byte line size
    Max CPUID level        0000001Bh
    Max CPUID ext. level    80000008h
    Cache descriptor    Level 1, D, 48 KB, 2 thread(s)
    Cache descriptor    Level 1, I, 32 KB, 2 thread(s)
    Cache descriptor    Level 2, U, 512 KB, 2 thread(s)
    Cache descriptor    Level 3, U, 6 MB, 16 thread(s)
    FID/VID Control        yes


    Turbo Mode        supported, enabled
    Max non-turbo ratio    12x
    Max turbo ratio        36x
    Max efficiency ratio    4x
    Speedshift        Autonomous
    O/C bins        none
    Ratio 1 core        36x
    Ratio 2 cores        36x
    Ratio 3 cores        33x
    Ratio 4 cores        33x
    Ratio 5 cores        33x
    Ratio 6 cores        33x
    Ratio 7 cores        33x
    Ratio 8 cores        33x
    IA Voltage Mode        PCU adaptive
    IA Voltage Offset    0 mV
    GT Voltage Mode        PCU adaptive
    GT Voltage Offset    0 mV
    LLC/Ring Voltage Mode    PCU adaptive
    LLC/Ring Voltage Offset    0 mV
    Agent Voltage Mode    PCU adaptive
    Agent Voltage Offset    0 mV
    TDP Level        15.0 W @ 10x
    TDP Level        12.0 W @ 7x
      TDP Level        25.0 W @ 12x

IFMA was introduced with Cannon Lake (i3-8121U) which is a 2 core cpu with HT, available in some NUCs, low speed low cost. https://en.wikipedia.org/wiki/Cannon...roarchitecture) https://en.wikipedia.org/wiki/AVX-512


What would be handy is a Windows version of lscpu, or an equivalent, which I've not been able to find. Sample Google colab !lscpu output

Architecture:        x86_64
CPU op-mode(s):      32-bit, 64-bit
Byte Order:          Little Endian
CPU(s):              2
On-line CPU(s) list: 0,1
Thread(s) per core:  2
Core(s) per socket:  1
Socket(s):           1
NUMA node(s):        1
Vendor ID:           GenuineIntel
CPU family:          6
Model:               63
Model name:          Intel(R) Xeon(R) CPU @ 2.30GHz
Stepping:            0
CPU MHz:             2300.000
BogoMIPS:            4600.00
Hypervisor vendor:   KVM
Virtualization type: full
L1d cache:           32K
L1i cache:           32K
L2 cache:            256K
L3 cache:            46080K
NUMA node0 CPU(s):   0,1
 Flags:               fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ss ht syscall nx pdpe1gb rdtscp lm constant_tsc rep_good nopl xtopology nonstop_tsc cpuid tsc_known_freq pni pclmulqdq ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm abm invpcid_single ssbd ibrs ibpb stibp fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid xsaveopt arat md_clear arch_capabilities

Installing WSL just to run lscpu seems like a lot of work. (Would really like to avoid system restarts on 24/7/365 boxes just for a brief utility run. lsgpu was ported. Maybe someone will port lscpu?)

In Windows, wmic cpu get * >wmic-cpu-get.txt provides a lot of info, but not such cpu flags.
For list formatting, wmic cpu list /format:list

[bsquared]

Ok, thanks. I take it you own that i5-1035G1? Does anyone know how to make windows report all cpu flags and instruction sets?

[Mark Rose]

I would boot a Linux live image and run `lscpu` in a terminal.

[kriesel]

Originally Posted by bsquared

Ok, thanks. I take it you own that i5-1035G1?

You're welcome. I picked up a used laptop because it had that 4-core&HT cpu supporting AVX512 and a nice big screen, and the cost was acceptable. It's running prime95 AVX512 code now.

Coreinfo does not list avx512 in example output at https://docs.microsoft.com/en-us/sys...loads/coreinfo. A quick test on the i5-1035G1 shows coreinfo's output does not include the strings avx512 or ifma there.

[bsquared]

Originally Posted by kriesel

In Windows, wmic cpu get * >wmic-cpu-get.txt provides a lot of info, but not such cpu flags.
For list formatting, wmic cpu list /format:list

Yes, wmic is not quite descriptive enough.

I just found "coreinfo", which is a free download from cnet, that reports flags. But maybe not all flags. For example on mine it reported AVX but not AVX2. This makes me doubt it will report all of the various subsets of AVX512...

[bsquared]

Originally Posted by kriesel

Coreinfo does not list avx512 in example output at https://docs.microsoft.com/en-us/sys...loads/coreinfo. A quick test on the i5-1035G1 shows coreinfo's output does not include the strings avx512 or ifma there.

Ok, thank you for checking.

Anandtech's review of ice lake and sunny cove says that it has IFMA52. And tiger lake / willow cove is supposed to be an optimized copy of sunny cove (with some cache enhancements). So processors from these families should be good to go. But some lscpu (or windows equivalent, whatever that is) assurance would be comforting.

[M344587487]

Speak of the devil: https://www.phoronix.com/scan.php?pa...ke-gcc11&num=1

[kruoli]

You can try y-cruncher. It will report something like this on startup (unfortunately, I have no AVX512, but this shows that y-cruncher should report it):

y-cruncher v0.7.8 Build 9503

Detecting Environment...

CPU Vendor:
    AMD         = Yes
    Intel       = No

OS Features:
  * 64-bit      = Yes
  * OS AVX      = Yes
  * OS AVX512   = No

Hardware Features:
    MMX         = Yes
  * x64         = Yes
  * ABM         = Yes
    RDRAND      = Yes
    RDSEED      = Yes
    BMI1        = Yes
  * BMI2        = Yes
  * ADX         = Yes
    MPX         = No
    PREFETCHW   = Yes
    PREFETCHWT1 = No
    RDPID       = Yes
    GFNI        = No
    VAES        = No

SIMD: 128-bit
  * SSE         = Yes
  * SSE2        = Yes
  * SSE3        = Yes
  * SSSE3       = Yes
    SSE4a       = Yes
  * SSE4.1      = Yes
  * SSE4.2      = Yes
    AES-NI      = Yes
    SHA         = Yes

SIMD: 256-bit
  * AVX         = Yes
    XOP         = No
  * FMA3        = Yes
  * FMA4        = No
  * AVX2        = Yes

SIMD: 512-bit
  * AVX512-F         = No
    AVX512-CD        = No
    AVX512-PF        = No
    AVX512-ER        = No
  * AVX512-VL        = No
  * AVX512-BW        = No
  * AVX512-DQ        = No
  * AVX512-IFMA      = No
  * AVX512-VBMI      = No

Alright Intel, how many drinks have you had tonight?
    AVX512-VPOPCNTDQ = No
    AVX512-4FMAPS    = No
    AVX512-4VNNIW    = No
    AVX512-VBMI2     = No
    AVX512-VPCLMUL   = No
    AVX512-VNNI      = No
    AVX512-BITALG    = No
    AVX512-BF16      = No


Auto-Selecting: 17-ZN1 ~ Yukina

N:\Kruse\Stresstests\y-cruncher v0.7.8.9503\Binaries/17-ZN1 ~ Yukina.exe


Launching y-cruncher...
================================================================


Checking processor/OS features...

Required Features:
    x64, ABM, BMI1, BMI2, ADX,
    SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2,
    AVX, FMA3, AVX2