Jump to content

Slovedon

Registered User
  • Posts

    97
  • Joined

  • Last visited

Posts posted by Slovedon

  1. 1 hour ago, bloodhawk said:

     

    First of all -

    DO NOT RUN FURMARK ON ANY GPU!!!!

     

     

    Thanks for the heads up. I never got that memo :/

     

    1 hour ago, bloodhawk said:

    Easiest way to eliminate any driver issues is to - Disconnect GPU, go into safe mode , use DDU to uninstall old drivers. When system is rebooting, connect the TB3 cable and once in Windows , install the new/drivers of your choice.

     

    Good to know, I'll give it a try if I have any driver problems in the future. I did a clean install of the desktop drivers, so hopefully that removed the original drivers completely

    • Thumbs Up 1
  2. Success! My Akitio Thunder2 + 980 setup is working fine (so far) with the 592G's Thunderbolt 3 connection and Akitio T3T adapter. Now I'm just waiting on Akitio to release the Node for full Thunderbolt 3 speeds 

     

    592G.jpg

     

    On 11/18/2016 at 0:28 PM, Andromedus said:

    That's excellent. If you don't mind, please post your results here (or a link to them). I'm very curious to know how the TB3 drivers respond to having a device plugged in. Does it start an update? Does it continue to read "no" for eGPU compatibility? Etc.

    Windows 10 will get confused with the 960M/980 combo and install the notebook and desktop Nvidia drivers repeatedly, so you have to disable device driver updates.

     

    The 980 showed in device manager after installing all the Acer driver & firmware updates from their website (most importantly the TB3 firmware) which let me install the desktop Nvidia drivers. After reboot, device manager gave a "OS could not find enough resources" error for the 980 (not uncommon when setting up eGPUs) but disabling, reenabling, and another reboot fixed it. Now the 960M and 980 are working perfectly for 10+ reboots, with and without eGPU connected.

     

    The eGPU compatibility section isn't showing for me, but like @bloodhawk said, that "no" is meaningless anyways

    • Thumbs Up 1
  3. I have the 15" version of this system ordered (Aspire V15 Nitro Black Edition aka Aspire VN7-592G) along with a Akitio T3T Thunderbolt 3 to Thunderbolt 2 adapter. Assuming that the T3T adapter works perfectly, I'll be able to test the 592G/792G's eGPU combatibility with my Akitio Thunder2 + GTX 980 setup

    • Thumbs Up 1
  4. Just now, Morv said:

     

    Ye I know, I'm just not into believing stuff. There's no proof whatsoever other than your word. No mentioning anywhere of this issue in the world wide web. If there is, please show it to me. Also how'd you know if this wasn't some issue that could be fixed with a firmware or driver update? The Core is out on the market for quite some time.

     

    I would much rather have them focus on making the simplest, most reliable, and most inexpensive TB3 graphics solution possible. Adding ports just adds more potential complications for the user and more potential scenarios for Akitio to test, no matter if those ports cause performance problems or not. It just doesn't seem worth it, especially when you can just use another hub to add more ports.

    • Thumbs Up 2
  5. Every powered riser I've used for my eGPU setup has had the following pin layout-

    hxyxQLA.png

    But the one I just received looks like this, with the black and yellow right next to each other-

    vuCVS3v.png

     

     

    Will this be an issue? I'm not sure if this changes how much power it can provide or not, so I thought I'd ask the experts before testing (don't want to risk damage, even though it probably wouldn't cause damage)

  6. On 2/27/2016 at 2:13 PM, Dschijn said:

    Yes there are adapters. But they are mini DisplayPort -> HDMI/VGA/DVI.

    Running a monitor on a TB port will result in a Optimus monitor -> not the best performance.

    For best performance the monitor has to be connected to the I/O of the eGPU.

     

    I'm actually looking to use a TB2 port as an input, so that I don't lose performance (which, as you say, happens with Optimus) when I'm using the internal LCD.

     

    On 2/28/2016 at 1:40 AM, Morv said:

    @Dschijn I think you didn't understand him. He wants to use the Thunderbolt connection as an input so the eGPU would connect directly to the internal display. I don't know what kind of miracles he's thinking about but that's what Optimus is doing...

     

    Yes, but we lose performance when using Optimus. If I can use Thunderbolt2 as an input, I can use the internal LCD with eGPU without the Optimus performance loss

  7. Is there any product out there that would allow us to use the second Thunderbolt 2 port on our Macbook Pro's as a video input? This would be a really nice boost to performance compared to using Optimus on the internal LCD. Even if there is minor input delay, I think it would still be worth it, considering the overall gaming experience is so much better when using an external monitor directly from GPU vs. using Optimus.

  8. I just completed this setup with the same 15" rMBP Iris Pro (i7-4750HQ) and Zotac 970 with the help of your guide. Thanks for posting.

    I'm actually glad my previously planned Asus G46vw Thunderbolt mobo upgrade didn't work out, because it let me find Apple's awesome refurbished offer, plus no more TB1. This is a helluva machine, and the ultimate eGPU system with 2xThunderbolt 2, and with that 1800p screen! Main downsides are non upgradeable CPU, and cooling isn't an issue (which it is for many) since we aren't using internal graphics. Hell, I may even use it with SLI eventually.

    How consistently does Optimus boot work for you? Are you turning it on before startup or at the boot menu? With my Zotac, I have to restart twice, and the second time occasionally fails for some reason, meaning 4x restarts to get Optimus going. tIt happens about 1 in 10 times, so not really much of a problem, but I might as well try to figure out why. I'm starting it up at the boot menu. I'm also on a standard Bootcamp setup of 8.1.

    Other than that, everything works perfectly fine for me, and the reference Zotac 970 gets a solid +210 core +110 memory overlock, while even boosting further past its stated limit.

  9. The motherboard from AliExpress got here, and unfortunately it isn't working. The listing pictured a rev 2.2 mobo that clearly showed the Thunderbolt chip, but they sent me a rev 2.0 without the chip - return in progress. Unless I can find a source that can guarantee a Thunderbolt mobo for a reasonable price (<$200) I may be holding off on an eGPU solution until I upgrade to a TB2 laptop.

  10. Are you able to use that without a powered riser? How does the Zotac card work compared to the recommended EVGA cards regarding optimus?

    Actually, I have not been able to test this setup yet. I'm waiting on my TB-enabled G46VW motherboard to arrive. It may very well need a powered riser, but I know that people have gotten a 760, 780 Ti, and other cards to work without a riser.

    Also, what do you mean by "recommended" EVGA cards?

  11. Based on your pics at http://forum.techinferno.com/enclosures-adapters/7205-us%24200-akitio-thunder2-pcie-box-16gbps-tb2-38.html#post106482 , can you remove the metal shroud over the fans? Looks to be four screws holding it down to the heatsink assembly. That would take off about 10mm on the right side. Surely the card would fit into the AKiTiO Thunder2 then?

    @Tech Inferno Fan @jacobsson Indeed it does! The shroud slides off easily after removing those 4 screws. You can see how the Akitio had to be bent slightly outward to accomodate the card with the shroud on, but it would have clearly fit comfortably without any alterations.

    post-14512-14494998511324_thumb.jpg

    post-14512-14494998510639_thumb.jpg

    • Thumbs Up 3
  12. All of these problems can be solved by replacing the outer case/cover with one both longer (2+ cm) and taller (1-2 cm), with a vent on the GPU side. The original end plates could be replaced by custom ones to close the gaps and maybe provide extra functionality (i.e. separate barrel connector for a Dell DA-2/other PSU, power switch, fan and LED mounts).

    They can also be fixed by not using the outer enclosure altogether :) That's my plan at least. I'll just use the outer enclosure for protecting the GPU while transporting it, since it slides in very easily, even with the slightly bent back plate to accommodate GPU length.

    - - - Updated - - -

    @jacobsson That's disappointing to hear. I'll be able to test and compare on my setup soon though. The TB mobo should be here in a week

    • Thumbs Up 1
  13. US$200+shipping AKiTiO Thunder2 for DIY eGPU users from [email protected]

    @Everyone, I've negotiated a special deal for our group. US$200/unit plus shipping AKiTiO Thunder2 purchased via [email protected]

    - Mention in your email that you are a DIY eGPU user

    - This will be a very good deal for US buyers

    - Asian customers will have their shipped from AKiTiO's Taiwan office and that should lower the shipping costs quite a bit.

    - European customers will probably be better off getting theirs from a local AKiTiO resellers due to high duty and shipping costs. eg: €169 mag.de, €249 macway.com (fr)

    And I thought my negotiating B&H from $270 to $240 was a good deal. Well done! It makes sense for them business-wise too, since that makes them the automatic choice for TB1 or TB2 enclosures in that price range.

    • Thumbs Up 1
  14. If we compare GPU scores to my TB1 @10Mbps, 'regular' clocked Zotac GTX 970:

    Core clock 1,076 MHz vs 1,314 MHz

    Memory bus clock 1,753 MHz vs 1,853 MHz

    3Dmark06:

    Result

    Total of 24% higher

    3Dmark Vant:

    Result

    21% higher GPU score

    3Dmark11:

    Result

    35% higher GPU score

    Fire Strike (11472 vs. 8872)

    29% higher GPU score

    I compared them wondering why there's that huge performance difference.

    I concluded the dfference is in your GPU's clock more than the TB1 versus TB2 bandwidth:

    Core clock 1,076 MHz vs 1,314 MHz (+22.1%)

    Memory bus clock 1,753 MHz vs 1,853 MHz (+5.7%)

    There may also be not-so-obvious power limits applied too that are further restricting performance.

    No sorry, I was planning to buy a wattmeter but since my egpu is stable it's not a priority anymore.

    If that can help you I tried 3dmark11 with your GPU's clock : NVIDIA GeForce GTX 970 video card benchmark result - Intel Core i7-4850HQ,Apple Inc. Mac-2BD1B31983FE1663

    So if we judge by that last result in 3DMark11, we can say that a GTX 970 on TB2 performs 18% faster than a GTX 970 on TB1. Although as Nando said, there could be other power limitations factoring in. And that's only a one test comparison.

    I'll be interested to see how my and other's 970s compare.

  15. Well, without a 2013 or newer MBP, you will not be able to just install windows and expect things to work. The hardware won't support it. Your only alternative on a 2012 rMBP is eGPU 1.3 - From what I understand, this will allow you to create a boot option that allows you to configure the PCIe interfaces. I takes some know how and that's why I'm glad I never had to mess with any of that.

    This is actually just a temporary testing system while I wait for a part for my main system that the setup is for (G46VW), and I'm not much of a Mac guy. I'm okay with TB1, 970 minus 15% performance is still very fast :)

    I'll try to get some advice from Nando for setup 1.3, although I've made progress. Connecting the box at the boot screen got it detected as a generic video device. I then installed the Nvidia drivers you recommended. Now it shows the Windows 7 boot screen on the external monitor for a split second and then goes black/freezes, and the internal goes black as soon as I select Windows boot.

×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use. We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.