Loading...
> >

Has anyone here bought a Treadripper rig and used it for Studio yet?

1356789

Comments

  • Robert FreiseRobert Freise Posts: 4,598

     

    ebergerly said:
    Robert Freise said:
    ebergerly said:

    Robert I'm curious...your choice of the Threadripper...was that because you have some apps that can utilize the cores, or was it more of a "future-proof" to make your rig last longer? 

    A bit of both

    Thanks, but I'm curious exactly what type of apps you have that might benefit from the many CPU cores of Threadripper. I keep seeing apps that seem to benefit much more from high power GPU's rather than CPU's, but companies keep coming out with multi-threaded CPU's. 

    Photography ,video conversion ,crompression and editing apps one example VideoRedo it'll use every cpu core as will VidCoder also C4d will benefit as it doesn't use the GPU

  • kyoto kidkyoto kid Posts: 41,925
    JamesJAB said:

    Well I was goung to comment about thread usage and dynamic cloth, but I was just suprised and dissapointed with Marvelous Designer...
    Daz Studio's Optitex Dynamic Cloth will use every single thread that you can throw at it (it will max out my 24 threads)
    Marvelous Designer on the other hand looks like it's locked down to 4 threads...

    ...yeah, but, MD isn't locked into content created with software none of us can afford.
  • JamesJABJamesJAB Posts: 1,766
    kyoto kid said:
    JamesJAB said:

    Well I was goung to comment about thread usage and dynamic cloth, but I was just suprised and dissapointed with Marvelous Designer...
    Daz Studio's Optitex Dynamic Cloth will use every single thread that you can throw at it (it will max out my 24 threads)
    Marvelous Designer on the other hand looks like it's locked down to 4 threads...

     

    ...yeah, but, MD isn't locked into content created with software none of us can afford.

    I guess the other part of that would be that MD still drapes much faster than the Daz Optitex...  4x CPU threads in MD is way faster than 24x CPU threads in Daz Optitex. surprise

  • kyoto kidkyoto kid Posts: 41,925

    ...yep.

  • GatorGator Posts: 1,320

    Bump, and @Robert Freise

    - have you put the system together yet and run Studio on it?

     

  • Robert FreiseRobert Freise Posts: 4,598
    scott762_948aec318a said:

    Bump, and @Robert Freise

    - have you put the system together yet and run Studio on it?

     

    Not yet the last of the parts will be arriving Tuesday Harvey caused a small delay

  • GatorGator Posts: 1,320
    Bummer. I was waiting hoping for Labor Day deals, I can tell you that you didn't miss anything not waiting. Itching to build myself.
  • Robert FreiseRobert Freise Posts: 4,598

    Still waiting for some parts received a notice that some coming from California were delayed because the trailer was late leaving then got another one saying it had been left off the trailer and was still at the dock and should be here Monday

    Most of the parts that were supposed to be here Tuesday are showing out for delivery today

     

  • GatorGator Posts: 1,320

    @Robert Freise

    Bummer, waiting sucks... especially after you paid for it!

    I just got a M2 NVM.e drive installed - decent upgrade over the SATA attached SSD drive.  Sped up booting, opening Word, Excel, and Photoshop quite a bit, but it's not that was long anyways.  A big scene with many Genesis 3 figures was taking 4 minutes to load, that dropped to 3 minutes 15 seconds.  Only a second or two saved saving the file. 

     

    BTW, piecing together a build.  What speed RAM did you go with?

  • Robert FreiseRobert Freise Posts: 4,598
    edited September 2017
    scott762_948aec318a said:

     

    BTW, piecing together a build.  What speed RAM did you go with?

    This is the ram I bought from NewEgg CORSAIR Dominator Platinum 128GB (8 x 16GB) 288-Pin DDR4 SDRAM DDR4 2800 (PC4 22400) Desktop Memory Model CMD128GX4M8B2800C14 

    At the time I ordered there were only five or six 128 GB sets available at NewEgg or Amazon and the fastest was out of my price range and this one was in the middle of the options

    There's an almost  complete list of my build on the second page of this thread only minor items missing from it

    Post edited by Robert Freise on
  • GatorGator Posts: 1,320
    Robert Freise said:
    scott762_948aec318a said:

     

    BTW, piecing together a build.  What speed RAM did you go with?

    This is the ram I bought from NewEgg CORSAIR Dominator Platinum 128GB (8 x 16GB) 288-Pin DDR4 SDRAM DDR4 2800 (PC4 22400) Desktop Memory Model CMD128GX4M8B2800C14 

    At the time I ordered there were only five or six 128 GB sets available at NewEgg or Amazon and the fastest was out of my price range and this one was in the middle of the options

    There's an almost  complete list of my build on the second page of this thread only minor items missing from it

    128 GB... woof, yeah that isn't cheap.  I am going with 64 GB, and that is pretty expensive to get the DDR4 3200 speed.  I am wondering how much performance gain there would be with that vs. slower memory.  I don't plan on overclocking the CPU much.

  • Robert FreiseRobert Freise Posts: 4,598
    scott762_948aec318a said:
    Robert Freise said:
    scott762_948aec318a said:

     

    BTW, piecing together a build.  What speed RAM did you go with?

    This is the ram I bought from NewEgg CORSAIR Dominator Platinum 128GB (8 x 16GB) 288-Pin DDR4 SDRAM DDR4 2800 (PC4 22400) Desktop Memory Model CMD128GX4M8B2800C14 

    At the time I ordered there were only five or six 128 GB sets available at NewEgg or Amazon and the fastest was out of my price range and this one was in the middle of the options

    There's an almost  complete list of my build on the second page of this thread only minor items missing from it

    128 GB... woof, yeah that isn't cheap.  I am going with 64 GB, and that is pretty expensive to get the DDR4 3200 speed.  I am wondering how much performance gain there would be with that vs. slower memory.  I don't plan on overclocking the CPU much.

    Yeah this is my Christmas and birthday present to myself so i splurged as much as I could.

    I've never done any overclocking and probably won't

    Don't have any idea how much of a performance gain there is between the ram speeds

  • ebergerlyebergerly Posts: 3,255

     

     

    scott762_948aec318a said:

    128 GB... woof, yeah that isn't cheap.  I am going with 64 GB, and that is pretty expensive to get the DDR4 3200 speed.  I am wondering how much performance gain there would be with that vs. slower memory.  I don't plan on overclocking the CPU much.

    I think that memory speed is yet another area where the "faster is better" hype doesn't necessarily apply. For some stuff it makes little to no difference, but it depends on stuff like what's the bottleneck in what you're doing at the time, and can your CPU handle more input from the memory or is it running a single thread that just can't take any more from the memory.

    And especially since anything above 3200 (?) is considered overclocking, you then have to worry about incompatibilities and BIOS revisions etc. Heck, the price difference between say 2133 and 3200 or higher might pay for a better GPU or CPU or storage, which might make a much bigger difference.

    Unfortunately, it depends. "You gotta pay to play" may not apply.  

  • ebergerlyebergerly Posts: 3,255

    BTW, I'm guessing that when it comes to DRAM speed when using Studio, the area that is important is the initial loading of the scene into memory, before it's transferred to the GPU VRAM. When I start an Iray render, it loads the scene, the DRAM utilization increases until the first Iray image appears in the render window. Then it flattens out while the Iray render is doing its thing. 

    So I'm guessing that maybe you might get some faster load times (assuming the CPU or storage isn't the limiting factor, which it might be), but once it's loaded into the GPU then the DRAM speed is relatively meaningless. 

    Now with other apps it might matter, though the videos I've seen on the subject lead me to believe that for graphics-intensive stuff you might be better off spending your money elsewhere. 

    But again, it depends on what you're doing. 

  • GatorGator Posts: 1,320
    ebergerly said:

     

     

    scott762_948aec318a said:

    128 GB... woof, yeah that isn't cheap.  I am going with 64 GB, and that is pretty expensive to get the DDR4 3200 speed.  I am wondering how much performance gain there would be with that vs. slower memory.  I don't plan on overclocking the CPU much.

    I think that memory speed is yet another area where the "faster is better" hype doesn't necessarily apply. For some stuff it makes little to no difference, but it depends on stuff like what's the bottleneck in what you're doing at the time, and can your CPU handle more input from the memory or is it running a single thread that just can't take any more from the memory.

    And especially since anything above 3200 (?) is considered overclocking, you then have to worry about incompatibilities and BIOS revisions etc. Heck, the price difference between say 2133 and 3200 or higher might pay for a better GPU or CPU or storage, which might make a much bigger difference.

    Unfortunately, it depends. "You gotta pay to play" may not apply.  

    Well, the answer is often "it depends", which is why I ask... I still gotta dig around, as Threadripper is so new I dunno.  It can make a substancial performance gain with some apps, others a small gain.

     

    A few posts above, I was pleasantly a little surprised to see the gain I saw loading a scene (few posts above) - A big scene with many Genesis 3 figures was taking 4 minutes to load, that dropped to 3 minutes 15 seconds.  Same system, only change is the SSD to M2 swap.

    Now, looking at standard performance counters that's difficult to see why I saw a substancial gain there - CPU is not nearly fully utilized, and the disk appeared to be hardly utilized at all.  So just looking at those numbers you would hardly expect any gain at all.  But shaving about 45 seconds off 4 minutes is a substancial gain.

  • Robert FreiseRobert Freise Posts: 4,598
    scott762_948aec318a said:
    ebergerly said:

     

     

    scott762_948aec318a said:

    128 GB... woof, yeah that isn't cheap.  I am going with 64 GB, and that is pretty expensive to get the DDR4 3200 speed.  I am wondering how much performance gain there would be with that vs. slower memory.  I don't plan on overclocking the CPU much.

    I think that memory speed is yet another area where the "faster is better" hype doesn't necessarily apply. For some stuff it makes little to no difference, but it depends on stuff like what's the bottleneck in what you're doing at the time, and can your CPU handle more input from the memory or is it running a single thread that just can't take any more from the memory.

    And especially since anything above 3200 (?) is considered overclocking, you then have to worry about incompatibilities and BIOS revisions etc. Heck, the price difference between say 2133 and 3200 or higher might pay for a better GPU or CPU or storage, which might make a much bigger difference.

    Unfortunately, it depends. "You gotta pay to play" may not apply.  

    Well, the answer is often "it depends", which is why I ask... I still gotta dig around, as Threadripper is so new I dunno.  It can make a substancial performance gain with some apps, others a small gain.

     

    A few posts above, I was pleasantly a little surprised to see the gain I saw loading a scene (few posts above) - A big scene with many Genesis 3 figures was taking 4 minutes to load, that dropped to 3 minutes 15 seconds.  Same system, only change is the SSD to M2 swap.

    Now, looking at standard performance counters that's difficult to see why I saw a substancial gain there - CPU is not nearly fully utilized, and the disk appeared to be hardly utilized at all.  So just looking at those numbers you would hardly expect any gain at all.  But shaving about 45 seconds off 4 minutes is a substancial gain.

    if  you went from a sata ssd to a M2 NVME (pci)  that's where your speed increase came from as the pci is faster than the sata

    If it was a M2 sata the difference would have to be in the chips used for the drive

  • GatorGator Posts: 1,320
    edited September 2017
    Robert Freise said:
    scott762_948aec318a said:
    ebergerly said:

     

     

    scott762_948aec318a said:

    128 GB... woof, yeah that isn't cheap.  I am going with 64 GB, and that is pretty expensive to get the DDR4 3200 speed.  I am wondering how much performance gain there would be with that vs. slower memory.  I don't plan on overclocking the CPU much.

    I think that memory speed is yet another area where the "faster is better" hype doesn't necessarily apply. For some stuff it makes little to no difference, but it depends on stuff like what's the bottleneck in what you're doing at the time, and can your CPU handle more input from the memory or is it running a single thread that just can't take any more from the memory.

    And especially since anything above 3200 (?) is considered overclocking, you then have to worry about incompatibilities and BIOS revisions etc. Heck, the price difference between say 2133 and 3200 or higher might pay for a better GPU or CPU or storage, which might make a much bigger difference.

    Unfortunately, it depends. "You gotta pay to play" may not apply.  

    Well, the answer is often "it depends", which is why I ask... I still gotta dig around, as Threadripper is so new I dunno.  It can make a substancial performance gain with some apps, others a small gain.

     

    A few posts above, I was pleasantly a little surprised to see the gain I saw loading a scene (few posts above) - A big scene with many Genesis 3 figures was taking 4 minutes to load, that dropped to 3 minutes 15 seconds.  Same system, only change is the SSD to M2 swap.

    Now, looking at standard performance counters that's difficult to see why I saw a substancial gain there - CPU is not nearly fully utilized, and the disk appeared to be hardly utilized at all.  So just looking at those numbers you would hardly expect any gain at all.  But shaving about 45 seconds off 4 minutes is a substancial gain.

    if  you went from a sata ssd to a M2 NVME (pci)  that's where your speed increase came from as the pci is faster than the sata

    If it was a M2 sata the difference would have to be in the chips used for the drive

    Yep, NVME drive... think same series you ordered, the Samsung 960 EVO. ETA: looked all over in my BIOS to check that it was running in x4 lanes but couldn't find a setting for it. Dunno if x1 or x4.
    Post edited by Gator on
  • Robert FreiseRobert Freise Posts: 4,598
    scott762_948aec318a said:
    Robert Freise said:
    scott762_948aec318a said:
    ebergerly said:

     

     

    scott762_948aec318a said:

    128 GB... woof, yeah that isn't cheap.  I am going with 64 GB, and that is pretty expensive to get the DDR4 3200 speed.  I am wondering how much performance gain there would be with that vs. slower memory.  I don't plan on overclocking the CPU much.

    I think that memory speed is yet another area where the "faster is better" hype doesn't necessarily apply. For some stuff it makes little to no difference, but it depends on stuff like what's the bottleneck in what you're doing at the time, and can your CPU handle more input from the memory or is it running a single thread that just can't take any more from the memory.

    And especially since anything above 3200 (?) is considered overclocking, you then have to worry about incompatibilities and BIOS revisions etc. Heck, the price difference between say 2133 and 3200 or higher might pay for a better GPU or CPU or storage, which might make a much bigger difference.

    Unfortunately, it depends. "You gotta pay to play" may not apply.  

    Well, the answer is often "it depends", which is why I ask... I still gotta dig around, as Threadripper is so new I dunno.  It can make a substancial performance gain with some apps, others a small gain.

     

    A few posts above, I was pleasantly a little surprised to see the gain I saw loading a scene (few posts above) - A big scene with many Genesis 3 figures was taking 4 minutes to load, that dropped to 3 minutes 15 seconds.  Same system, only change is the SSD to M2 swap.

    Now, looking at standard performance counters that's difficult to see why I saw a substancial gain there - CPU is not nearly fully utilized, and the disk appeared to be hardly utilized at all.  So just looking at those numbers you would hardly expect any gain at all.  But shaving about 45 seconds off 4 minutes is a substancial gain.

    if  you went from a sata ssd to a M2 NVME (pci)  that's where your speed increase came from as the pci is faster than the sata

    If it was a M2 sata the difference would have to be in the chips used for the drive

     

    Yep, NVME drive... think same series you ordered, the Samsung 960 EVO. ETA: looked all over in my BIOS to check that it was running in x4 lanes but couldn't find a setting for it. Dunno if x1 or x4.

    Motherboard specs should give you the info on the M2 slot

  • nicsttnicstt Posts: 11,715

    @OP

    Yes bought, but awaiting delivery.

  • Robert FreiseRobert Freise Posts: 4,598
    edited September 2017

    As what I'm waiting on are some brackets and fittings I've been looking at fan and radiator positioning and I'm giving consideration to getting another one of the cases and using it for the liquid cooling as that would let me positon the radiators where they would both be pulling air in from outside for more efficient cooling.

    There are some options for that with just the one case but for a front mount I would have to do away with the optical drive which I don't want to do as I still believe in burning backup files to disk for safe keeping off site as well as on site or for a bottom mount I wold have to loose one sey of drive cages so I could mount the PS one that side

    Post edited by Robert Freise on
  • GatorGator Posts: 1,320
    scott762_948aec318a said:
    Robert Freise said:
    scott762_948aec318a said:
    ebergerly said:

     

     

    scott762_948aec318a said:

    128 GB... woof, yeah that isn't cheap.  I am going with 64 GB, and that is pretty expensive to get the DDR4 3200 speed.  I am wondering how much performance gain there would be with that vs. slower memory.  I don't plan on overclocking the CPU much.

    I think that memory speed is yet another area where the "faster is better" hype doesn't necessarily apply. For some stuff it makes little to no difference, but it depends on stuff like what's the bottleneck in what you're doing at the time, and can your CPU handle more input from the memory or is it running a single thread that just can't take any more from the memory.

    And especially since anything above 3200 (?) is considered overclocking, you then have to worry about incompatibilities and BIOS revisions etc. Heck, the price difference between say 2133 and 3200 or higher might pay for a better GPU or CPU or storage, which might make a much bigger difference.

    Unfortunately, it depends. "You gotta pay to play" may not apply.  

    Well, the answer is often "it depends", which is why I ask... I still gotta dig around, as Threadripper is so new I dunno.  It can make a substancial performance gain with some apps, others a small gain.

     

    A few posts above, I was pleasantly a little surprised to see the gain I saw loading a scene (few posts above) - A big scene with many Genesis 3 figures was taking 4 minutes to load, that dropped to 3 minutes 15 seconds.  Same system, only change is the SSD to M2 swap.

    Now, looking at standard performance counters that's difficult to see why I saw a substancial gain there - CPU is not nearly fully utilized, and the disk appeared to be hardly utilized at all.  So just looking at those numbers you would hardly expect any gain at all.  But shaving about 45 seconds off 4 minutes is a substancial gain.

    if  you went from a sata ssd to a M2 NVME (pci)  that's where your speed increase came from as the pci is faster than the sata

    If it was a M2 sata the difference would have to be in the chips used for the drive

     

    Yep, NVME drive... think same series you ordered, the Samsung 960 EVO.

     

    scott762_948aec318a said:
    Robert Freise said:
    scott762_948aec318a said:
    ebergerly said:

     

     

    scott762_948aec318a said:

    128 GB... woof, yeah that isn't cheap.  I am going with 64 GB, and that is pretty expensive to get the DDR4 3200 speed.  I am wondering how much performance gain there would be with that vs. slower memory.  I don't plan on overclocking the CPU much.

    I think that memory speed is yet another area where the "faster is better" hype doesn't necessarily apply. For some stuff it makes little to no difference, but it depends on stuff like what's the bottleneck in what you're doing at the time, and can your CPU handle more input from the memory or is it running a single thread that just can't take any more from the memory.

    And especially since anything above 3200 (?) is considered overclocking, you then have to worry about incompatibilities and BIOS revisions etc. Heck, the price difference between say 2133 and 3200 or higher might pay for a better GPU or CPU or storage, which might make a much bigger difference.

    Unfortunately, it depends. "You gotta pay to play" may not apply.  

    Well, the answer is often "it depends", which is why I ask... I still gotta dig around, as Threadripper is so new I dunno.  It can make a substancial performance gain with some apps, others a small gain.

     

    A few posts above, I was pleasantly a little surprised to see the gain I saw loading a scene (few posts above) - A big scene with many Genesis 3 figures was taking 4 minutes to load, that dropped to 3 minutes 15 seconds.  Same system, only change is the SSD to M2 swap.

    Now, looking at standard performance counters that's difficult to see why I saw a substancial gain there - CPU is not nearly fully utilized, and the disk appeared to be hardly utilized at all.  So just looking at those numbers you would hardly expect any gain at all.  But shaving about 45 seconds off 4 minutes is a substancial gain.

    if  you went from a sata ssd to a M2 NVME (pci)  that's where your speed increase came from as the pci is faster than the sata

    If it was a M2 sata the difference would have to be in the chips used for the drive

     

    Yep, NVME drive... think same series you ordered, the Samsung 960 EVO.

     

    Robert Freise said:
    scott762_948aec318a said:
    Robert Freise said:
    scott762_948aec318a said:
    ebergerly said:

     

     

    scott762_948aec318a said:

    128 GB... woof, yeah that isn't cheap.  I am going with 64 GB, and that is pretty expensive to get the DDR4 3200 speed.  I am wondering how much performance gain there would be with that vs. slower memory.  I don't plan on overclocking the CPU much.

    I think that memory speed is yet another area where the "faster is better" hype doesn't necessarily apply. For some stuff it makes little to no difference, but it depends on stuff like what's the bottleneck in what you're doing at the time, and can your CPU handle more input from the memory or is it running a single thread that just can't take any more from the memory.

    And especially since anything above 3200 (?) is considered overclocking, you then have to worry about incompatibilities and BIOS revisions etc. Heck, the price difference between say 2133 and 3200 or higher might pay for a better GPU or CPU or storage, which might make a much bigger difference.

    Unfortunately, it depends. "You gotta pay to play" may not apply.  

    Well, the answer is often "it depends", which is why I ask... I still gotta dig around, as Threadripper is so new I dunno.  It can make a substancial performance gain with some apps, others a small gain.

     

    A few posts above, I was pleasantly a little surprised to see the gain I saw loading a scene (few posts above) - A big scene with many Genesis 3 figures was taking 4 minutes to load, that dropped to 3 minutes 15 seconds.  Same system, only change is the SSD to M2 swap.

    Now, looking at standard performance counters that's difficult to see why I saw a substancial gain there - CPU is not nearly fully utilized, and the disk appeared to be hardly utilized at all.  So just looking at those numbers you would hardly expect any gain at all.  But shaving about 45 seconds off 4 minutes is a substancial gain.

    if  you went from a sata ssd to a M2 NVME (pci)  that's where your speed increase came from as the pci is faster than the sata

    If it was a M2 sata the difference would have to be in the chips used for the drive

     

    Yep, NVME drive... think same series you ordered, the Samsung 960 EVO. ETA: looked all over in my BIOS to check that it was running in x4 lanes but couldn't find a setting for it. Dunno if x1 or x4.

    Motherboard specs should give you the info on the M2 slot

    Documentation says it supports PCIEx4 and SATA mode...  I can't find where the setting is.  frown

  • Robert FreiseRobert Freise Posts: 4,598
    scott762_948aec318a said:
    scott762_948aec318a said:
    Robert Freise said:
    scott762_948aec318a said:
    ebergerly said:

     

     

    scott762_948aec318a said:

    128 GB... woof, yeah that isn't cheap.  I am going with 64 GB, and that is pretty expensive to get the DDR4 3200 speed.  I am wondering how much performance gain there would be with that vs. slower memory.  I don't plan on overclocking the CPU much.

    I think that memory speed is yet another area where the "faster is better" hype doesn't necessarily apply. For some stuff it makes little to no difference, but it depends on stuff like what's the bottleneck in what you're doing at the time, and can your CPU handle more input from the memory or is it running a single thread that just can't take any more from the memory.

    And especially since anything above 3200 (?) is considered overclocking, you then have to worry about incompatibilities and BIOS revisions etc. Heck, the price difference between say 2133 and 3200 or higher might pay for a better GPU or CPU or storage, which might make a much bigger difference.

    Unfortunately, it depends. "You gotta pay to play" may not apply.  

    Well, the answer is often "it depends", which is why I ask... I still gotta dig around, as Threadripper is so new I dunno.  It can make a substancial performance gain with some apps, others a small gain.

     

    A few posts above, I was pleasantly a little surprised to see the gain I saw loading a scene (few posts above) - A big scene with many Genesis 3 figures was taking 4 minutes to load, that dropped to 3 minutes 15 seconds.  Same system, only change is the SSD to M2 swap.

    Now, looking at standard performance counters that's difficult to see why I saw a substancial gain there - CPU is not nearly fully utilized, and the disk appeared to be hardly utilized at all.  So just looking at those numbers you would hardly expect any gain at all.  But shaving about 45 seconds off 4 minutes is a substancial gain.

    if  you went from a sata ssd to a M2 NVME (pci)  that's where your speed increase came from as the pci is faster than the sata

    If it was a M2 sata the difference would have to be in the chips used for the drive

     

    Yep, NVME drive... think same series you ordered, the Samsung 960 EVO.

     

    scott762_948aec318a said:
    Robert Freise said:
    scott762_948aec318a said:
    ebergerly said:

     

     

    scott762_948aec318a said:

    128 GB... woof, yeah that isn't cheap.  I am going with 64 GB, and that is pretty expensive to get the DDR4 3200 speed.  I am wondering how much performance gain there would be with that vs. slower memory.  I don't plan on overclocking the CPU much.

    I think that memory speed is yet another area where the "faster is better" hype doesn't necessarily apply. For some stuff it makes little to no difference, but it depends on stuff like what's the bottleneck in what you're doing at the time, and can your CPU handle more input from the memory or is it running a single thread that just can't take any more from the memory.

    And especially since anything above 3200 (?) is considered overclocking, you then have to worry about incompatibilities and BIOS revisions etc. Heck, the price difference between say 2133 and 3200 or higher might pay for a better GPU or CPU or storage, which might make a much bigger difference.

    Unfortunately, it depends. "You gotta pay to play" may not apply.  

    Well, the answer is often "it depends", which is why I ask... I still gotta dig around, as Threadripper is so new I dunno.  It can make a substancial performance gain with some apps, others a small gain.

     

    A few posts above, I was pleasantly a little surprised to see the gain I saw loading a scene (few posts above) - A big scene with many Genesis 3 figures was taking 4 minutes to load, that dropped to 3 minutes 15 seconds.  Same system, only change is the SSD to M2 swap.

    Now, looking at standard performance counters that's difficult to see why I saw a substancial gain there - CPU is not nearly fully utilized, and the disk appeared to be hardly utilized at all.  So just looking at those numbers you would hardly expect any gain at all.  But shaving about 45 seconds off 4 minutes is a substancial gain.

    if  you went from a sata ssd to a M2 NVME (pci)  that's where your speed increase came from as the pci is faster than the sata

    If it was a M2 sata the difference would have to be in the chips used for the drive

     

    Yep, NVME drive... think same series you ordered, the Samsung 960 EVO.

     

    Robert Freise said:
    scott762_948aec318a said:
    Robert Freise said:
    scott762_948aec318a said:
    ebergerly said:

     

     

    scott762_948aec318a said:

    128 GB... woof, yeah that isn't cheap.  I am going with 64 GB, and that is pretty expensive to get the DDR4 3200 speed.  I am wondering how much performance gain there would be with that vs. slower memory.  I don't plan on overclocking the CPU much.

    I think that memory speed is yet another area where the "faster is better" hype doesn't necessarily apply. For some stuff it makes little to no difference, but it depends on stuff like what's the bottleneck in what you're doing at the time, and can your CPU handle more input from the memory or is it running a single thread that just can't take any more from the memory.

    And especially since anything above 3200 (?) is considered overclocking, you then have to worry about incompatibilities and BIOS revisions etc. Heck, the price difference between say 2133 and 3200 or higher might pay for a better GPU or CPU or storage, which might make a much bigger difference.

    Unfortunately, it depends. "You gotta pay to play" may not apply.  

    Well, the answer is often "it depends", which is why I ask... I still gotta dig around, as Threadripper is so new I dunno.  It can make a substancial performance gain with some apps, others a small gain.

     

    A few posts above, I was pleasantly a little surprised to see the gain I saw loading a scene (few posts above) - A big scene with many Genesis 3 figures was taking 4 minutes to load, that dropped to 3 minutes 15 seconds.  Same system, only change is the SSD to M2 swap.

    Now, looking at standard performance counters that's difficult to see why I saw a substancial gain there - CPU is not nearly fully utilized, and the disk appeared to be hardly utilized at all.  So just looking at those numbers you would hardly expect any gain at all.  But shaving about 45 seconds off 4 minutes is a substancial gain.

    if  you went from a sata ssd to a M2 NVME (pci)  that's where your speed increase came from as the pci is faster than the sata

    If it was a M2 sata the difference would have to be in the chips used for the drive

     

    Yep, NVME drive... think same series you ordered, the Samsung 960 EVO. ETA: looked all over in my BIOS to check that it was running in x4 lanes but couldn't find a setting for it. Dunno if x1 or x4.

    Motherboard specs should give you the info on the M2 slot

    Documentation says it supports PCIEx4 and SATA mode...  I can't find where the setting is.  frown

    This may be helpful http://www.tomshardware.com/answers/id-3011699/sata-ssd-pcie-ssd.html

  • kyoto kidkyoto kid Posts: 41,925
    ebergerly said:

     

     

    scott762_948aec318a said:

    128 GB... woof, yeah that isn't cheap.  I am going with 64 GB, and that is pretty expensive to get the DDR4 3200 speed.  I am wondering how much performance gain there would be with that vs. slower memory.  I don't plan on overclocking the CPU much.

    I think that memory speed is yet another area where the "faster is better" hype doesn't necessarily apply. For some stuff it makes little to no difference, but it depends on stuff like what's the bottleneck in what you're doing at the time, and can your CPU handle more input from the memory or is it running a single thread that just can't take any more from the memory.

    And especially since anything above 3200 (?) is considered overclocking, you then have to worry about incompatibilities and BIOS revisions etc. Heck, the price difference between say 2133 and 3200 or higher might pay for a better GPU or CPU or storage, which might make a much bigger difference.

    Unfortunately, it depends. "You gotta pay to play" may not apply.  

    ....indeed. I have read on tech sites that for many operations there isn't even much of a performance differnece between DDR3 and DDR4. The downside with DDR3 is the largest kit is 64 GB and most is proprietary to a specific brand of workstation like Dell or HP, not a "whitebox" system like I plan to build.

  • kyoto kidkyoto kid Posts: 41,925
    edited September 2017

    ...so another question.  Has there been word of anyone who successfully got Threadripper to work with W7.  As I understand there is a way to do so with the Ryzen CPUs (somewhat complicated but aparently succesful) which Threadripper is based off of?

    Post edited by kyoto kid on
  • GatorGator Posts: 1,320
    edited September 2017
    Robert Freise said:

    As what I'm waiting on are some brackets and fittings I've been looking at fan and radiator positioning and I'm giving consideration to getting another one of the cases and using it for the liquid cooling as that would let me positon the radiators where they would both be pulling air in from outside for more efficient cooling.

    There are some options for that with just the one case but for a front mount I would have to do away with the optical drive which I don't want to do as I still believe in burning backup files to disk for safe keeping off site as well as on site or for a bottom mount I wold have to loose one sey of drive cages so I could mount the PS one that side

    I totally lost faith in discs as I found out the hard way the dye degrades on them... I had some go bad.  They were protected from light too, they were stored in binders.  Don't use them anymore myself, although I think they may have gotten better for long term storage.

    Have you thought about a slim external optical drive?

    Post edited by Gator on
  • Robert FreiseRobert Freise Posts: 4,598

    I haven't tried mainly because I only have one Win7 license and it's in use

  • Robert FreiseRobert Freise Posts: 4,598
    scott762_948aec318a said:
    Robert Freise said:

    As what I'm waiting on are some brackets and fittings I've been looking at fan and radiator positioning and I'm giving consideration to getting another one of the cases and using it for the liquid cooling as that would let me positon the radiators where they would both be pulling air in from outside for more efficient cooling.

    There are some options for that with just the one case but for a front mount I would have to do away with the optical drive which I don't want to do as I still believe in burning backup files to disk for safe keeping off site as well as on site or for a bottom mount I wold have to loose one sey of drive cages so I could mount the PS one that side

    I totally lost faith in discs as I found out the hard way the dye degrades on them... I had some go bad.  They were protected from light too, they were stored in binders.  Don't use them anymore myself, although I think they may have gotten better for long term storage.

    Have you thought about a slim external optical drive?

     

    Yeah thought about that but still wouldn't give me the needed clearance

  • Robert FreiseRobert Freise Posts: 4,598

    Finally getting started my living room looks like a computer shop

    Here's the chasis 

    MonsterCase.jpg
    2268 x 4032 - 2M
  • GatorGator Posts: 1,320

    Wow, looks like a monster setup there!  Appears that you got two of those Thermaltake Core x9 cases?

    Also looks like the PC parts on top, water cooling on bottom?  May want to look into that, that's naturally what I would do but my AIO coolers say they radiators should be above the device.

    Also, forgot a piece of info from a friend heavy into water cooling, he recommends against any sort of dyed fluid.  He said one UV dye separated on him, was quite a PITA as he had to disassemble the whole setup.  IIRC a simple flush wasn't sufficient to clean up the blocks too, or he needed some special stuff to get the blocks clean.  Long story short a lot of wasted time.

  • ebergerlyebergerly Posts: 3,255

    So do you guys really get that much of a performance boost to justify overclocking and the expense of water cooling? I mean, if you're going from say 3.5 GHz to 4.0 GHz by overclocking, does that really translate to a significant performance difference in what you're doing? I can't imagine it matters much with Studio, especially for rendering. 

    Or are you like me where you just like playing with cool new technology? smiley

Sign In or Register to comment.