Jump to content

kaladeth

Registered User
  • Posts

    17
  • Joined

  • Last visited

Posts posted by kaladeth

  1. I'm waiting for my new HP ProBook 640 to arrive (I know - it's already a year old!). It only comes with a dual core processor, so from both (a) thermal structure and (B) BIOS point-of-view, can the HP ProBook 640 handle a (47W or 57W) Quad Core processor?

    The service manual says the max CPU you can put in the HP ProBook 640 is an Intel Dual Core i7-4600M 2.80-GHz processor (3.0-MB L3 cache, 37 W) 737330-001:

    http://h20628.www2.hp.com/km-ext/kmcsdirect/emr_na-c04030196-1.pdf

    But from a physical socket point of view, I can put in an i7-4940MX:

    post-19997-14494999233299_thumb.jpg

    Is the heat from the increased TDP the only problem? Or will the BIOS throttle the CPU down to 37W?

  2. In large part due to this community (kudos!), I just recently purchased an 8570p and stuck an i7-3940XM in it just for kicks! :) I did a quick check with my current eGPU setup and it works PnP! I'll be redoing the 3DMark benchmarks soon, but in the meantime I wanted to post my initial Throttlestop numbers for your comments (and my newbie questions):

    Table 1. VID and TDP vs. Multiplier

    post-19997-14494998956983_thumb.png

    I tried running this initially with a 65W power supply, but it only let me get to 45 TDP. Switching to a 90W power supply fixed that problem.

    Questions:

    1. My x12 VID and TDP seem to be significantly higher than the others on the tracking board. Is this because the 8570P generally uses more power vs. the 2570P?

    2. Why does my VID jump to 1.2 (and the clock jumps to 3.7GHz) when I try any multiplier between 30 and 36?

    3. Why does my clock not change from 3.7GHz when I try a multiplier of 38 and 39? Is there a different software that I need to use? Or do I need to change anything in the BIOS?

    4. Does RAM affect any of these tests?

  3. I'm thinking of upgrading my base hardware (from an HP ProBook 6460b) to an HP ProBook 640 G1 Notebook PC. Three questions:

    1. I checked the quick specs, and it says it has a Smart Card reader, but no mention of an Express Card slot. I checked the service guide and it does mention ExpressCard, but has "(depending on the configuration)" in the component title. Can anyone confirm if this laptop model supports PCMCIA / ExpressCard for our eGPUs?

    Quick Specs: HP ProBook 640 G1 Notebook PC HP ProBook 650 G1 Notebook PC (QuickSpecs/c04126035.pdf)

    Data Sheet: http://h71016.www7.hp.com/dstore/html/pdfs/US_CA_English_HP_ProBook_640_Datasheet.pdf

    Service Guide: http://h20628.www2.hp.com/km-ext/kmcsdirect/emr_na-c04030196-1.pdf

    2. From a value POV, maybe the i7-4700MQ is the best CPU for this, but am curious to know what's the most powerful chip that I can put in this laptop? Will an i7-4910MQ work (Max TDP = 47W)?

    PassMark - Intel Core i7-4910MQ @ 2.90GHz - Price performance comparison

    3. I noticed the Haswell chips support PCI Express 3.0. Is this backwards compatible with our Express 2.0 eGPU cards?

  4. Quote

    Thank you very much kaladeth! :) Did you get those results during the TS-bench? If that's the case, you did it right! It would be a great find for SB systems, because Tech Inferno Fan's findings here would appear to be limited to his specific system (HP 2560P).



    @noric - I didn't run TS Bench sorry... I reviewed nando's test procedure and did the test properly this time (I hope!). Here's the full table, including temperature Max:

    x12 x23 x24 x25 x26 x27 x28 x29 x30 x31 x32 x33 x34 x35
    i7-2860QM VID 0.8206 0.9707 0.9857 1.0007 1.0308 1.0558 1.0858 1.1158 XX XX XX XX XX XX
    45W / kaladeth TDP 14 27.8 29.4 31.1 33.5 37.6 41.2 44.2 XX XX XX XX XX XX
    TEMP 68 73 78 80 83 86 90 93


    When I try to go past x29 the power goes past 45W for a few cycles and the CPU gets throttled back down to ~x28.6.

    @Nando - Do you think running with a better heat sink / without the back cover will let the 2860QM go past x29? Would you know if it can reach x32 theoretically?
    • Thumbs Up 1
  5. @kaladeth

    Nice to see such a nice improvement with a change in CPU. My i7-2620M is only a smidgen better than the i5-2520M you had. You are tempting me to upgrade to one of the nicer i7 CPUs lol.

    I see you have 3 displays. Have you done testing with 3 hooked up vs 1 hooked up? I noticed a pretty solid performance difference with a change in the number of displays active:

    http://forum.techinferno.com/diy-e-gpu-projects/2109-diy-egpu-experiences-%5Bversion-2-0%5D-43.html#post74494

    (1 to 6 displays in that test)

    @angerthosenear - Go for it! :) I think I tried testing with only 1 display hooked up before and didn't see a big difference. Too lazy to test again hehe.

  6. Hey nando, what about idle power consumption in SB vs. IB? That's a very important factor in a mobile environment. Did you say anything about IB being more consuming at idle? This could only be tested on IB system, by pulling the IB cpu and plugging a similar specifications SB cpu...

    @kaladeth: could you please try to see in throttlestop the voltage and TDP at maximum multithread performance and at x27 multiplier (such as nando is testing)? Also, what about the temps of your 2860qm? Thanks.

    @noric - Sorry for the delay! Not sure if I used the program correctly, but the voltage is around ~1.23V and ~42.3W when set to Turbo. When set to x27 multiplier, the voltage drops to ~1.06V and ~24.1W.

    post-19997-14494997106129_thumb.jpg

    • Thumbs Up 1
  7. What do You mean about microeGPU? What is connection between PCI-E generation and smaller eGPU? :)

    I prefer MSI TwinFrozr or DirectCUII than Windfroce, I used to have maaaany GPU's but windforce is very good system but so fake plastic ;) DirectCUII and TwinFrozr are the coolest and the quitest as possible :)

    Take a look that You high end CPU from SB is the same level (GS in 3dm11) like my TDP limited IB CPU (according to @Tech Inferno Fan results) :)

    @bjorm - I was thinking if the card can be powered through the bus, then we wouldn't need the external power supply, so the physical space taken by the egpu can be effectively halved. I'm with you on the DirectCUII :).

  8. Tech Inferno Fan>> Full hardware spec of eGPU implementation posted earlier : http://forum.techinferno.com/diy-e-gpu-projects/2109-diy-egpu-experiences-%5Bversion-2-0%5D-222.html#post75247

    HP ProBook 6460b (with Intel HD Graphics 3000)

    Intel Core i5-2520M

    EVGA GeForce GTX 670 2GB

    PE4L ver.2.1b

    PCIe 2.0 enabled in the BIOS

    @lapytopy - Thanks for sharing your case pix. That looks really portable!

    @bjorm - If the next gen PE4L lets us power through the bus (is this really theoretically possible?) then you can make a micro eGPU :)

    @noric - GTX 670 for sure... Windforce if you can get your hands on one. If I had the chance to start over I would have held out for the Asus GTX 670 DirectCU II (or maybe the Asus GTX 670 DC Mini!)

    I was finally able to upgrade my humble i5-2520M CPU to an i7-2860QM. I still can't touch any of the leaderboard numbers, but I'm happy with the boost. Here are some before and after snapshots:

    3DMARK06

    17268 - NVIDIA GeForce GTX 670 video card benchmark result - Intel Core i5-2520M Processor,Hewlett-Packard 161C

    23128 - NVIDIA GeForce GTX 670 video card benchmark result - Intel Core i7-2860QM Processor,Hewlett-Packard 161C

    3DMARK11

    P5942 - Graphics Score 7382, Physics Score 3764

    P7155 - Graphics Score 7427, Physics Score 7083 - NVIDIA GeForce GTX 670 video card benchmark result - Intel Core i7-2860QM Processor,Hewlett-Packard 161C

    I didn't need to do anything to the heatsink or fan, but I went ahead and upgraded to a 90W laptop power supply. Honestly I don't think I really needed to though.

    • Thumbs Up 3
  9. ... I should also mention that I got a new CPU...

    So what did you get! I am thinking of sticking a 2960XM in but wondering if I will melt the board... :nightmare:. Has anyone ever tried putting a 55 Max TDP CPU in a laptop chassis that's supposed to be rated for 45 Max TDP only? I understand I need to get a bigger power supply, and do fun stuff with the heat sink and fan, but wondering if the delivery system can take it.

    Btw, I uploaded a few pics of my setup (I updated my original post - link in my sig). I think I need a larger desk now haha.

  10. Ok I got it working! The bootup sequence is tricky (boot computer, pause boot, switch on eGPU, wait for D1 to switch off, continue boot)... if I deviate from that process, the boot process doesn't continue or, the eGPU is not found. Otherwise, there were no problems with memory allocation.

    My setup again:

    HP ProBook 6460b (with Intel HD Graphics 3000)

    Intel Core i5-2520M

    EVGA GeForce GTX 670 2GB

    PE4L ver.2.1b

    PCIe 2.0 enabled in the BIOS

    3dm06 = 17268 NVIDIA GeForce GTX 670 video card benchmark result - Intel Core i5-2520M Processor,Hewlett-Packard 161C

    3dm11 = P5942 NVIDIA GeForce GTX 670 video card benchmark result - Intel Core i5-2520M Processor,Hewlett-Packard 161C

    post-19997-14494996630567_thumb.jpg

    Here are a few more pictures. I stuck everything in a SilverStone Sugo SG02F case (still plenty of room in there!), I then used it as a stand for my laptop, and hid the whole mess behind my center monitor. Game on!

    post-19997-14494996672365_thumb.jpg

    I did have a few questions:

    1. The benchmark scores seem a tad low (above figures are for external monitor) compared to what you guys are getting... did I miss a step?

    2. I'm using the latest Nvidia drivers, so am I already at x1.2Opt?

    3. Is x1.2Opt just for Fermi cards or can the Fermi drivers also work for Kepler cards?

    4. Does driving 3 external monitors affect the benchmarks (I see it running on only 1 screen).

    5. Lastly, I can't seem to switch off Nvidia Surround after I've switched it on... has anyone encountered this too?

    Appreciate any tips!

    post-19997-14494996670206_thumb.jpg

    post-19997-14494996671305_thumb.jpg

  11. Thanks Nando. That makes sense since it has twice the cores/threads.

    Now since the OP was asking in the context of an eGPU implementation, I grabbed a few samples from the leaderboards:

    DX9

    15" Lenovo T520 i7-2860QM 2.5 16 [email protected] 23749 226.8 Kuro-D & & QM67 EC Win7/64
    12" Lenovo X220/T i5-2520M 2.5 8 [email protected] 18078 144 arsenal243 rhx123 QM67 EC Win7/64


    DX10

    15" Lenovo T520 i7-2860QM 2.5 16 [email protected] 29463 295.3 Kuro-D & & QM67 EC Win7/64
    12" Lenovo X220/XT i5-2520M 2.5 8 [email protected] 25847 197.4 arsenal243 rhx123 QM67 EC Win7/64


    DX11

    15" Lenovo T520 i7-2860QM 2.5 16 [email protected] 8626 2441 Kuro-D & & QM67 EC Win7/64
    12" Lenovo X220/T i5-2520M 2.5 8 [email protected] 8585 2347 arsenal243 rhx123 QM67 EC Win7/64


    So if I upgraded from an i5-2520M to a an i7-2860QM, from an eGPU perspective it seems a Quad can deliver ~31% improvement on DX9, ~14% on DX10, and <0.5% on DX11.

    Great advice on the heat/cooling consideration... will share back findings here when I get to Phase 2 (CPU upgrade) :).
  12. Hi guys! First of all thank you to Nando and the rest of the community for keeping this scene alive. It's encouraged me to go ahead and try an eGPU build of my own! I haven't built up a PC since my 486 all those many years ago (!) so doing the research in preparation for this brought back so many memories.

    My planned setup: 14" HP ProBook 6460b i5-2520M @ 2.50GHz 16.0 GB, 3x ViewSonic VX2453, EVGA GeForce GTX 670, DIY eGPU (PE4L-EC060A), Corsair CX600M, SilverStone SG02B-F

    I thought briefly of upgrading my motherboard to an 8460p and upgrading my CPU to a Quad Core, but thought I'd try it with the 6460b first and see how far the Dual Core i5 can go. Any other 6460b owners out there?

    Components are on the way hopefully in time for a nice gaming session this weekend! I'll post up a pic once it's up and running, but in the meantime wish me luck!

×
×
  • 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.