Waiting for Mac Pro 7,1

Discussion in 'Mac Pro' started by Mago, May 31, 2016.

Tags:
  1. Mago thread starter macrumors 68000

    Joined:
    Aug 16, 2011
    Location:
    Beyond the Thunderdome
    #6426
    Much better if they keep the price while reducing costs
    So the iPhone Plus should also responsible for the entire iPad-line decline? .... I dont think are related.
    So you disqualifies yourself as you admit you are not an iPad (whichever size) user, there are iPad specific users that prefer having something smaller, as Pilots (you cant run Jeppepsen Flite Deck on an iPhone, or Garmin Pilot, you need an flull fledged iPad, otherwise the FAA disqualifies you) they preffer lighter/smaller.

    This is my own personal prediction about:, at sometime Apple will launch the iPad Pro mini, at 8" but having exactly the same prize and features as the 10" iPad Pro, sharing all the same features including pricing tiers everything except size and weight.. [insert Tim Cook voice ] Amazing... [/insert Tim Cook's voice].
     
  2. AidenShaw macrumors P6

    AidenShaw

    Joined:
    Feb 8, 2003
    Location:
    The Peninsula
    #6427
    AMD doesn't support SGX, which has been in the last few x64 chips.
     
  3. Mago thread starter macrumors 68000

    Joined:
    Aug 16, 2011
    Location:
    Beyond the Thunderdome
    #6428
    For Good...

    KabyLake supported Extensions:
    MOVBE, MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, FSGSBASE, RDRND, FMA3, F16C, BMI, BMI2, VT-x, VT-d, TXT, TSX, RDSEED, ADCX, PREFETCHW, CLFLUSHOPT, XSAVE, SGX, MPX

    SGX,MPX are used in DRM related Applications, VT- are used by virtualization software as Hipervisor, Virtual Machine.

    FSGSBASE,TXT,TSX I dont remenber where are is used but it where Present on the previous AMD cpu (excavator) so AMD decided to discontinue it.

    It just means Zen CPU are not as good as intel for VMWare (most VM do not need VT-), and some DRM apps wont run on Zen (this is good).

    Zen Supported Extensions:
    MOVBE, MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, POPCNT, AVX, AVX2, AES, PCLMUL, RDRND, F16C, BMI, BMI2, RDSEED, ADCX, PREFETCHW, CLFLUSHOPT, XSAVE, SHA, CLZERO

    SHA
    is a criptograpy extension also to be adopted by next Intel architecture .

    CLZERO
    is an new ISA instruction introduced to prevent some families of exploits, not clear about Intel Adopting it in the future.
     
  4. antonis, Oct 1, 2017
    Last edited: Oct 1, 2017

    antonis macrumors 68000

    antonis

    Joined:
    Jun 10, 2011
    #6429
    IIRC the reason was the inability/reluctancy of IBM to provide a G5 cpu for the mac laptops. Hence, mac laptops got stuck with the G4 for too long leading to a big performance gap with the competition. Apple had really left with no other choice but switch.
     
  5. Mago thread starter macrumors 68000

    Joined:
    Aug 16, 2011
    Location:
    Beyond the Thunderdome
    #6430
  6. CWallace macrumors 603

    CWallace

    Joined:
    Aug 17, 2007
    Location:
    Seattle, WA
    #6431
    While nVidia is writing compatible drivers, third-party hacks to allow them to be used via eGPUs won't be of any benefit if Apple chooses to use a custom GPU card in the MP 7,1.
     
  7. Peace macrumors Core

    Peace

    Joined:
    Apr 1, 2005
    Location:
    Space--The ONLY Frontier
    #6432
  8. AidenShaw macrumors P6

    AidenShaw

    Joined:
    Feb 8, 2003
    Location:
    The Peninsula
    #6433
    AMD supports AMD-v (aka SVM) which is roughly equivalent to VT-x. It also has EPT (https://en.wikipedia.org/wiki/X86_virtualization#AMD_virtualization_.28AMD-V.29).

    VT-x (or equivalents) enable huge performance improvements for most VM workloads, and has been part of almost every CPU released since 2011 (and in many before that).

    SGX is also useful for secure apps not involving DRM. For example, an SGX-enabled banking application can't leak information to kernel malware. The app encrypts its RAM footprint - so not even kernel code that can look at the RAM contents can see the actual data.

    Not having SGX is not a good thing.
     
  9. Mago thread starter macrumors 68000

    Joined:
    Aug 16, 2011
    Location:
    Beyond the Thunderdome
    #6434
    AMD enables full RAM encryption and AES extenxions, it is better than SGX enclaves AES encrypt ranges of RAM an app w/o the key to access some range wont be capable to do nothing.
    --- Post Merged, Oct 1, 2017 ---
    eGPU in HS depends on Apple enababling nVidia so its support its beta for Apple engineers so the can test those drivers and later include in a future update Tesla's HW ID.
     
  10. Mactrunk macrumors regular

    Joined:
    May 12, 2005
    #6435
    Just revisiting this thread and I'm encouraged to hear that there might be a 7,1.

    I salute the long time experts here who are keeping this thread alive.
    I'm humbled by their immense knowledge and depth.
    I hope Apple is listening to you!

    I'm still on a 5,1... waiting for a revelation.
     
  11. ventricle macrumors member

    ventricle

    Joined:
    Nov 28, 2016
    Location:
    Oz
    #6436
    Lucky you. I returned from holiday to find a red LED lighting up my 5,1. I think it’s dead. I thought I had unplugged it but maybe I didn’t. With a H’tosh I’ve lost 15% functionality but gained 100% in performance. I plan on giving that away if the 7,1 qualifies as a 5,1 replacement by my standards. I too have been waiting. For way too long.
     
  12. goMac macrumors 603

    Joined:
    Apr 15, 2004
    #6437
    Lot's of "it depends" here.

    If Nvidia updates their drivers to be compatible with eGPU, shouldn't be a problem. Apple doesn't really have a reason to block them, they support all sorts of third party drivers over Thunderbolt. An Nvidia card is just another type of card over Thunderbolt.

    If Nvidia starts going through the Made for Mac certification program again you might even see Apple start rolling those drivers into macOS again. But that's less important with eGPU in since it's not like eGPU will keep your machine from booting without the right drivers.

    And the big unknown with the 7,1 is if Apple will allow third party GPUs internally, which could also mean the Nvidia drivers for newer cards could get bundled back into the OS.

    Apple's biggest problem is they don't have any Nvidia hardware for their compatibility labs, because there isn't any officially blessed Nvidia Mac hardware coming out of Nvidia. Can't test every new release on a certified-for-Mac 1080 Ti if there isn't one....
     
  13. ActionableMango macrumors 604

    ActionableMango

    Joined:
    Sep 21, 2010
    #6438
    I keep seeing this mentioned as a huge problem, but historical examples have proven that no loopback cable is required. In fact, no special hardware of any kind was required to use discreet GPU's output from an integrated GPU's physical port.

    Using a software solution you can mirror the discrete GPU's framebuffer on the integrated GPU's framebuffer. This gives you the power of the discreet GPU while outputting the video signal from the integrated GPU's Thunderbolt port.

    The first cases where I've seen people doing this are laptops where someone had hooked up an eGPU via expresscard-based PCIe. Naturally a loopback cable cannot be used in this scenario because almost no laptops have a graphics-in port. How this works was that the eGPU handled the rendering while the laptop's screen handled the output via the laptop's iGPU by copying the eGPU's framebuffer. There was third party software drivers that accomplished this on Windows, but being third party it was problematic. Even with the very limited PCIe available through an expresscard slot, people were turning laptops into gaming machines via eGPU.

    The far more famous and working software was the Nvidia Optimus Copy Engine. From Anandtech:

    ...the GPU does the necessary work, and the final result is then copied from the GPU frame buffer into the IGP frame buffer over the PCI Express bus.

    With Optimus, all of the video connections come through the IGP, so there's no extra hardware on the motherboard.
     
  14. ManuelGomes, Oct 4, 2017
    Last edited: Oct 4, 2017

    ManuelGomes macrumors 65816

    Joined:
    Dec 4, 2014
    Location:
    Aveiro, Portugal
    #6439
  15. Joe The Dragon macrumors 6502

    Joined:
    Jul 26, 2006
    #6440
    but the workstation hardware in the mac pro does not have any igpu and they don't have an IPMI+basic video card as well. The PRO HP workstations have TB cards (pci-e+MB header that needs an DP loop back from the video card)
     
  16. ActionableMango macrumors 604

    ActionableMango

    Joined:
    Sep 21, 2010
    #6441
    Yes, but an onboard GPU can be added as part of any newly designed 7,1 and the TB ports can be attached to that. I'm not suggesting that this is a solution for existing Mac Pro models.
     
  17. askunk macrumors regular

    askunk

    Joined:
    Oct 12, 2011
    Location:
    London
    #6442
    I don't see the point for a TB3 eGPU for a Mac Pro, the internal bus should be expected to be much faster than 40 Gpbs. Polaris cards are already limited by the TB3 transfer rate.

    As far as the internal GPU is concerned. DP is native on TB3, it will all be solved with a dongle, Apple style. I'm pretty sure you will be able to connect the TBD as well as DP displays. What I will hope for is that Apple is going to ask vendors to actually produce a dedicated card with a TB3 output.

    We could fall back into the situation where there will be only a few cards available, at higher prices than PC cards like with the cMP. If it has to be modular, I expect the mMP to be able to adopt any PCI card on the market, as long as drivers are available. (therefore NVidia, too)
     
  18. vailr macrumors member

    vailr

    Joined:
    Oct 22, 2009
    #6443
  19. Mago thread starter macrumors 68000

    Joined:
    Aug 16, 2011
    Location:
    Beyond the Thunderdome
    #6444
    it is for providing DP1.3 feed and maybe USB3.1 data to usb-c only displays, not a thunderbolt related solution, neither provides alt modes, just DP1.3 over USB-C, and maybe USB3.1 signal data.
     
  20. CWallace macrumors 603

    CWallace

    Joined:
    Aug 17, 2007
    Location:
    Seattle, WA
    #6445
    Yeah, LG makes a line of computer displays that can use USB-C for their video input (along with more traditional connectors) so I am guessing that is what MSI is aiming for with adding the connector to the 11G.
     
  21. ManuelGomes macrumors 65816

    Joined:
    Dec 4, 2014
    Location:
    Aveiro, Portugal
    #6446
    Soon enough we'll get Titan Ridge with DP1.4 support.
     
  22. CWallace macrumors 603

    CWallace

    Joined:
    Aug 17, 2007
    Location:
    Seattle, WA
    #6447
    Yup. I expect that will be 80 gigabits per second to allow 8K at 60Hz and 5K at 120Hz.
     
  23. pat500000 macrumors 604

    pat500000

    Joined:
    Jun 3, 2015
    #6448
    How are nvidia web driver and cuda update working with osx hs?
     
  24. singhs.apps macrumors regular

    Joined:
    Oct 27, 2016
    #6449
    8k at 60hz would require around 50 Gbps data to work. Not sure thunderbolt 3 can handle that, let alone usb 3.1 gen 2.

    Maybe two tb 3 cables.

    Only single cable option would be HDMI 2.1
     
  25. CWallace macrumors 603

    CWallace

    Joined:
    Aug 17, 2007
    Location:
    Seattle, WA
    #6450
    TB3 only supports DP 1.2 and 8K needs DP 1.3 for 30Hz or 1.4 for 60Hz so TB3 can't drive an 8K display, period.

    I expect Apple's rumored 8K display, if it is launched, will use HDMI 2.1 and the new Mac Pro will have the same.
     

Share This Page