Jump to content

YuuKi-AnS

Member
  • Posts

    45
  • Joined

  • Last visited

Posts posted by YuuKi-AnS

  1. On 2/16/2022 at 9:27 PM, Binary_Nexus said:

    So, I've tried flashing that old BIOS to my card and now I've got something different from what I had before.

    Attempting to start the card with micctrl --start echos "The Intel(R) Xeon Phi(TM) coprocessor is starting. Error manipulating coprocessors: Error starting device."

    So instead I tried resetting the card using micctrl -r and nothing happens so I view the status of the card with micctrl -s and it echos the error "mic0: reset failed (mode: N/A image: C:\Program Files\Intel\MPSS\filesystem\bzImage-knightscorner.bin)", which I have never had before. With BIOS 1.0.0.1159 it would issue the node 0 reset command to the card correctly but starting the card would cause it to time out after attempting to boot from the knightscorner bzimage. I feel like the older BIOS is either confusing the card or MPSS (or both), and as 1.0.0.1140 is shorter compared to 1.0.0.1159 there may have been some slightly different configs or bugs that the newer version corrected?


    Which card did you get version 1.0.0.1140 from by the way? Was it Knights Ferry Rev. F or something earlier? Because if it's the same board revision as my card (Cherrypeak Fab. F) then that could mean that there are some smaller revisions for each major revision of A through F (i.e. small BIOS alterations, presence of debug ports, extra or fewer components, etc.), something which I hadn't really considered before.

     

     

    GPU=ED

    Board=Fab F HPC

    GPU=D0

    Number:*122928*

  2. On 2/16/2022 at 9:27 PM, Binary_Nexus said:

    So, I've tried flashing that old BIOS to my card and now I've got something different from what I had before.

    Attempting to start the card with micctrl --start echos "The Intel(R) Xeon Phi(TM) coprocessor is starting. Error manipulating coprocessors: Error starting device."

    So instead I tried resetting the card using micctrl -r and nothing happens so I view the status of the card with micctrl -s and it echos the error "mic0: reset failed (mode: N/A image: C:\Program Files\Intel\MPSS\filesystem\bzImage-knightscorner.bin)", which I have never had before. With BIOS 1.0.0.1159 it would issue the node 0 reset command to the card correctly but starting the card would cause it to time out after attempting to boot from the knightscorner bzimage. I feel like the older BIOS is either confusing the card or MPSS (or both), and as 1.0.0.1140 is shorter compared to 1.0.0.1159 there may have been some slightly different configs or bugs that the newer version corrected?


    Which card did you get version 1.0.0.1140 from by the way? Was it Knights Ferry Rev. F or something earlier? Because if it's the same board revision as my card (Cherrypeak Fab. F) then that could mean that there are some smaller revisions for each major revision of A through F (i.e. small BIOS alterations, presence of debug ports, extra or fewer components, etc.), something which I hadn't really considered before.

     

    They all seem to be called CherryPeak CRB FAB:F…

    85ACB047-5712-47AA-ABFF-1BEC0AF06B29.jpeg

    EE22E4E3-D89D-4612-B742-39DC83E9EA97.jpeg

  3. On 1/25/2022 at 10:26 AM, erek said:

    yes, there's at least one other known in the wild.  it's in Russia

     

    still missing drivers to run 3D.

     

    not promoting the listing, but a follow up to the Linus Video, historical reference and awareness only

    We're starting at KnightFerry trying to find a way to start Larrabee.😶

    The auction price for this Larrabee on eBay I think is way too expensive...🥲

    But I'm still looking forward to a collector who bought this Larrabee willing to talk to us about this card and try to get it started.😶

    FC7F3E80-9038-4281-A8B1-AEBD891BE919.jpeg

    9C1CD3B7-B9A7-4014-8052-DA0638969BC9.jpeg

    1EC2C90B-30C6-45D4-BB18-6A005B7C82E0.png

    F71F925C-F225-455A-8444-368C8405D6BC.png

  4. On 7/27/2021 at 7:19 PM, Carl Johnson W16 said:

    Today in another forum i got the screenshots for a certain "fake" GPU, but it seems like it is indeed legit and completely one off, im predicting that it is some sort of internal early development card that got slapped with a cooler and sold. Adding the GPU-Z screenshot below. 

    Linus, if you are reading this, i think this is going to be in @YuuKi-AnS territory.

     

    Ekran Görüntüsü (4).png

    If you want to confirm whether it is an engineering sample, can you send us a photo of this card for confirmation...?

  5. Hey guys, what do you think of this processor.

    F64B5EB2-0763-40CD-9439-CFBD3F983ECB.thumb.jpeg.965d0daeec9b497e25eb911ff359aa43.jpeg

    -

    intel 4th Gen Xeon Processor

     Code name:

    Sapphire Rapids

     Q-sepc code: QVV5

     Frequency: 1.30GHz

    (normal value of early samples)

     Socket: LGA4677-X1FF44400-BC32-4CC6-8A94-08BDEED6CE80.thumb.jpeg.0a66cb575e8108289a4eb62db57ec1ab.jpeg

     Version: ES0 late/ES1

     Internal heat conduction: brazing

     Multi-element: Yes

    (MAX up to 4 HBM + 4 Core)

    Process technology:10nm+++ ESF

     New application support for main hardware:

    DDR5 RAM / PCIe 5.0 / onboard HBM

    「This one of mine may be a low-end model without HBM inside and I found an FPGA chip in this processor」

    E36C7C95-9EC0-4241-A197-B41B815D3FBF.thumb.jpeg.260da8ee74dd3c8b8f82da020cab7817.jpeg

    175B2109-F74E-44BD-9CE3-C0B707290288.thumb.jpeg.d99a934d8fce418a96fb1608d2de2544.jpeg

    Its core is very similar to the Microsoft logo.

    89B936FF-4991-42D3-B72D-3ECB1759A00D.thumb.jpeg.3565c0ba93f7abc6a703c154da804f46.jpeg

    22AEB219-4242-4975-A028-8B404326E56F.jpeg

    DE907C70-0E61-4994-9857-B199DB83E2D1.jpeg

  6. 4 hours ago, porina said:

    Looking at the R15 score of 2454, need something to compare against.

     

    I did have a previous run of 3700X fixed at 4.3 GHz, scoring 2238. I don't have 8 core Intel to compare with. Nearest I have is 6 core 8086k OC to 5.2 scoring 1740. This would scale to a score 2320 at 8 cores. Of course, scaling wont be exact, but Cinebench in general is very scalable. As a double check, my past architecture testing showed Zen 2 had over 20% IPC increase over Skylake in R15. The difference here is only 17% so might be some mild signs of real world scaling coming into play as core counts increase. Anyway, reason for mentioning this is that on average Rocket Lake is expected to have around 18% IPC increase over Skylake, as that is what's claimed for the similar architecture Ice Lake mobile. While I wont attempt to put an exact number to it, that would imply the 11900k might have been running in the low 4.x GHz during R15, which is a reasonable all-core turbo.

     

    @YuuKi-AnS are you able to state the clock while running R15? Can I assume your sample is QS? (Don't have much experience, but I find ES is less likely to report the full CPU name)

    Yes, this is a QS sample, it runs this turbo frequency of 5.2GHz.

  7. The third-generation Intel Xeon processor is divided into two architectures,

    one is CooperLake, the other is iceLake. We have tested Intel's iceLake 10nm sample.

    34FEFC08@80B01513.7296B65F_recompress.jpg.66b0de4d5b119fe2423f1020d711371e.jpg

    This is related information:

    intel 3rd Gen iceLake-SP Xeon CPU

     Q-sepc: QSTN-A0 (EngineeringSample)

     Architecture: IceLake-SP "Server"

     Intel official release date: 2021.Q1

     Interface: LGA4189-5

     Process technology: intel 10nm

     Forecast series: intel Xeon Silver 3300 Series

     Core/Thread: 14 cores / 28 threads

     Cache: L1: 1.1MB / L2: 17.5MB / L3: 21.0MB

     Reference frequency: 2.00/2.20GHz (1995MHz)

     Maximum Turbo Frequency: 4.00GHz 

     TDP: 165w Maximum temperature: 105℃

     SMT/CMP: 2/14

     Current step/revision: A0 / 0

     [The voltage is not 1.6v, AIDA64 does not know this CPU, and CPUz does not know, many of them are empty]

    1050707081_@O8AJLK5IWEC2@SDBZKG.png.550921feea815681fc3b64183f1d2aa2.png37FCFC0B@05D89F05.7296B65F_recompress.jpg.9c0b4636a7b210e26ac3538fcba7e1ec.jpg

    ---

    wh.jpg.c0959c27e77b3dd1297682abf5bc3e16.jpg

    Test platform: intel Whitley LGA4189-5 motherboard
     (EngineeringSample-ES/QS)
    cb20.jpg.1085ac9d45683fbba94df4d4f963fb32.jpg
    CineBench R20: 5,900
     CPU-z v1.94.8: 371.6 / 6363 (V17.01)
     CPU-z v1.94.8: 120.5/ 1965.8 (V19.01)
     CPU-z v1.94.8: 553.1/10038.4 (AVX-512 single shot)
     Ferritz chess: 19,715 [16 threads]
    c.jpg.b1caafc877eb95cbd0382b5e7e547fcc.jpg
  8. 5 minutes ago, Kilrah said:

    Really, an Intel CPU plonked on a random card and everything censored makes it an "engineering sample" now?

    What a load of BS and awful job of attempting to grab attention...

    It hasn’t been released yet, and the manufacturer doesn’t allow me to reveal many things.

    I can't disclose the back of the GPU core either.

  9. Bilibili:结城安穗-YuuKi_AnS
    YuuKi-AnS 结城安穗 (China)

    NVIDIA RTX 3090  GPU NVIDIA RTX3090 GPU

    3×8pin power supply, 12-phase power supply, 24GB video memory (12 particles on both sides)
    PCIe x16

    A manufacturer’s top non-public card OEM RTX 3090 ES EngineeringSample

    0D481BDA5C327D3F9B41F8ED66A439D8.thumb.png.12d62e86e626b8b3a1213e365e3e54f3.png

    F4EC5F2FE951D14C2582B4714348FBAF.png.e126a5fa66360f4e72b42e365843501d.png

    Because of confidentiality, I will not release the complete picture, and I don't want to lose my job.

     

  10. Hey guys, did you seen this fake GPU(xe)?

    IMG_0927.thumb.png.b180b3d43051925706ddd543020a20cb.png

    IMG_0926.thumb.png.cf6b8efda87b14aac4d726b14068b80f.png

     Some days ago, I get a Xeon Phi Co-processor on XianYu (Like ebay), when I get it and open its cover, I was so surprised, here are its pictures.

    (I know Xeon Phi is not GPU)

    IMG_0900.thumb.jpeg.e93ab872439fabe6930fa2d1021e94eb.jpegIMG_0902.thumb.jpeg.bcfca880e8dbb8c2462239f1f27b3c8b.jpegIMG_0905.thumb.jpeg.046abb6577d22a6c5834937522ef7c6c.jpegIMG_0920.thumb.jpeg.a542f694a0e8b599cca0a4549a488abf.jpegIMG_0958.thumb.jpeg.3cccf93e11cb15af82c1d952d57803ed.jpeg

    Its core is same as Intel Xeon Phi x200 family CPU. But its BGA.

    This Xeon Phi has a PLX:PEX8733 and intel chipset C612J.   It has a MicroUSB to test.

    I don't know why, intel often make some so strange EngineeringSample like this.

    AIDA64:

    IMG_0981.thumb.png.78eaec11121c8647f7e06dab24f387cb.png

    CPU World:

    IMG_1175.thumb.png.d0cb03f674b3fb74c84b3bef9432b7b0.pngIMG_1176.jpeg.62cacc148ee1620b734406235a034585.jpeg

    This is my video link:https://www.bilibili.com/video/BV1za4y1a7rj/

    Bilibili:結城安穗-YuuKi_AnS

    (My English is not good, may have some mistakes)

  11. 6 hours ago, Binary_Nexus said:

    My card doesn't show up as a display adapter for me. Just shows up in Device Manager as "coprocessor", not under display adapters

    GPU Z and Speccy both show nothing for me

    In cmd I can use micinfo and it just says "Knights Ferry - unknown error"

     

    Which exact driver did you use for the card and how exactly did you install it?

    I just install MPSS 3.8.6 just normal installation.

  12. 2 minutes ago, Binary_Nexus said:

    Which version of MPSS did you use? My card didn't work with MPSS at all. The card uOS got stuck in a boot loop when I tried to start the card with MPSS commands as it was trying to boot from a Knights Corner image rather than Knights Ferry.

    3.3.4 and 3.8.6

×