Daz Studio Iray - Rendering Hardware Benchmarking

1323335373845

Comments

  • JB007JB007 Posts: 118

    The Vertex Doctor said:

    gsil247 said

    I'm strongly considering getting the 4090, and yet I have the 3090. Already been planning how I'm going to get the funds, so by the holidays I should be able to afford it. By then official benchmarks will come out and that will help me decide if its worth it or not. That is probably the only good thing about them being so expensive; time to save and time to think.

    Same here. Already planning out the specs for the system I will build next year. CPU isn't as important to me so will probably get a current gen i5 since I am currently running an i5-6600K BUT I do intend to get the 4090 as well as a beefy SeaSonic 1600W PSU to handle pairing it with my current 3090 and NVLink. Outervision PSU calculator comes out with 1219 W recommended so a 1600W Seaosnic Platinum should do the trick. The room I run my PC in has a circuit that can handle a total of 2300W so I should be able to get by.

    I don't think there's an NVLink port on the 4090 ...

  • PerttiAPerttiA Posts: 10,024

    JB007 said:

    The Vertex Doctor said:

    gsil247 said

    I'm strongly considering getting the 4090, and yet I have the 3090. Already been planning how I'm going to get the funds, so by the holidays I should be able to afford it. By then official benchmarks will come out and that will help me decide if its worth it or not. That is probably the only good thing about them being so expensive; time to save and time to think.

    Same here. Already planning out the specs for the system I will build next year. CPU isn't as important to me so will probably get a current gen i5 since I am currently running an i5-6600K BUT I do intend to get the 4090 as well as a beefy SeaSonic 1600W PSU to handle pairing it with my current 3090 and NVLink. Outervision PSU calculator comes out with 1219 W recommended so a 1600W Seaosnic Platinum should do the trick. The room I run my PC in has a circuit that can handle a total of 2300W so I should be able to get by.

    I don't think there's an NVLink port on the 4090 ...

    No, there isn't and even if there was, one could only link identical cards. 

  • skyeshotsskyeshots Posts: 149

    3DMinh said:

    What is the best version of Daz Studio for rendering speed? I have the previous versions of daz studio, but I don't know which to install. Is version 4.16.0.3 better than 4.15 ? Is from 4.16 to 4.20 the rendering speed starting to slow down?

    Thank you.

    Such a good question.

    In September, I posted a guess of 28 iterations per second for the 4090 at the time of release. From johndoe_36eb90b0 bench scores, Daz 4.16 ran at 31.79 iterations per second and Daz 4.21 was 22.30 iterations per second. If we take the average of these two versions, we are at 27.045. This would put my guess at 97% accuracy!

    Honestly though, the GPU scores tend to be overwhelmingly consistent regardless of peripheral hardware, since Daz rendering is mostly GPU bound. For the 4090 (from johndoe_36eb90b0 samples) we are seeing a 30% drop in performance between 4.16 > 4.21 on the 4090 baseline here. For my A6000 cards, I would say that this is a similar trend.

  • fred9803fred9803 Posts: 1,564

    JB007 said:

    I don't think there's an NVLink port on the 4090 ...

    On the 4090 PCB there are traces and contacts for the NVLink connector but they are under the solder mask. Perhaps a legacy from the 3090 or NVIDIA intended an NVLink for the 4090 but later thought better of it.... probably a wise move.

  • skyeshotsskyeshots Posts: 149

    System/Motherboard: SuperMicro X12
    CPU: 2x Xeon Gold 6348 @ Stock 3.5 GHZ
    GPU: A6000 (95% Power + 1150 Mem)
    System Memory: 512 GB ECC @ 3200 mhz 
    OS Drive: WD SN850 1 TB NVMe
    Asset Drive: 256 GB RAM DRIVE 
    PSU: Corsair AX1600i
    Operating System: Win 11 Pro
    Nvidia Drivers Version: 516.59
    DAZ Studio 4.21 (64-bit) Public Build +BETA+
     
    GPU Only:
    2022-10-23 18:04:00.952 Iray [INFO] - IRAY:RENDER ::   1.0   IRAY   rend info : Device statistics:
    2022-10-23 18:04:00.952 Iray [INFO] - IRAY:RENDER ::   1.0   IRAY   rend info : CUDA device 0 (NVIDIA RTX A6000): 1800 iterations, 2.159s init, 119.639s render
    2022-10-23 18:04:01.724 [INFO] :: Finished Rendering
    2022-10-23 18:04:01.773 [INFO] :: Total Rendering Time: 2 minutes 5.35 seconds
    Rendering Performance: 1800/119.639 = 15.045 Iterations Per Second
    Loading Time: 5.711 Seconds

    With CPUs
    2022-10-23 17:56:06.445 Iray [INFO] - IRAY:RENDER ::   1.0   IRAY   rend info : Device statistics:
    2022-10-23 17:56:06.445 Iray [INFO] - IRAY:RENDER ::   1.0   IRAY   rend info : CUDA device 0 (NVIDIA RTX A6000): 1588 iterations, 2.099s init, 106.252s render
    2022-10-23 17:56:06.445 Iray [INFO] - IRAY:RENDER ::   1.0   IRAY   rend info : CPU:                              212 iterations, 0.999s init, 105.774s render
    2022-10-23 17:56:07.173 [INFO] :: Finished Rendering
    2022-10-23 17:56:07.217 [INFO] :: Total Rendering Time: 1 minutes 51.78 seconds
    Rendering Performance: 1800/106.252 = 16.94 Iterations Per Second
    Loading Time: 6.006 Seconds

    Showing 12% faster w/CPU enabled today. Usually the opposite happens. Perhaps some enhancements have arrived for CPU/GPU usage? 

  • The Vertex DoctorThe Vertex Doctor Posts: 198
    edited October 2022

    JB007 said:

    I don't think there's an NVLink port on the 4090 ...

    Yeah, I just found that out. WTH man? NVlink that finally allowed you to combine memory between cards? Ridiculous. 

    Post edited by The Vertex Doctor on
  • JB007JB007 Posts: 118
    edited October 2022

    Is it fair to say then that the 4090 is working in Daz ... just not with all the taps turned on yet? Faster than the 3090 in Iray .. but not just by much yet?

    I've got my 4090 system ready to be built .. but I don't want to go for it just yet and be sitting around for 4 to 6 weeks (potentially without dForce) with something that's just a little bit faster than what I have with my 3090 right now .. 

    So is the 4090 cooked or should I wait until the oven is turned up first?

     

    Post edited by JB007 on
  • KeikuKeiku Posts: 143

    NVLINK would be pretty useless with the behemoth size of new GPUs. You couldn't actually fit them together.

    Quickly testing Gen 9 on my RTX 3090. They are noticably slower than Gen 8 even without any additional morphs. I'm wondering if plunking down another $1800 for a 4090 will help to offset this? I'd hate to spend that much again for a GPU so soon. But if it helps, I may just do it.

    No additional morphs were added yet to Gen 9 or Vicky 9 other than what came in the initial packages.

    Base Genesis took 13 seconds to load.
    Regular Vicky took 20 seconds.
    HD Vicky took 22 seconds.

    But Genesis 8.1 with a bunch of morphs took 8 seconds to load.
    Victoria 8 HD took 10 seconds.

    So Gen 9 is 1.62x slower than 8.1 and Vicky 9 HD is 2.2x slower than Vicky 8 HD, again without adding any morph packages to 9.

  • PerttiAPerttiA Posts: 10,024

    Character loading times under 30 seconds are nothing. When people are talking about slow loading times, it's usually measured in tens of minutes, loading times of 3-5 minutes are pretty good when one has a lot of content for the figure.

  • KeikuKeiku Posts: 143
    edited October 2022

    PerttiA said:

    Character loading times under 30 seconds are nothing. When people are talking about slow loading times, it's usually measured in tens of minutes, loading times of 3-5 minutes are pretty good when one has a lot of content for the figure.

    But I'm spoiled.

    I do wonder however, if this would still scale linearly. Would people with inferior hardware or significantly more morphs experience the same relative slow down? I'm curious what the load times will be once I do add a bunch of Gen 9 content that rivals what I have for 8.1. If it's already up to 2x slower, I may indeed see multi-minute load times.

    Post edited by Keiku on
  • System Configuration
    System/Motherboard
    : Asus tuf x670e-plus
    CPU: AMD Ryzen 9 7950x g @ 4.5 GHz (default)
    GPU: Gigabyte rtx 4090 Gameing OC 24g @ 2535 MHz (default)
    System Memory: CORSAIR Dominator Platinum 288-Pin DDR5 64G @ 5200
    OS Drive: SAMSUNG 980 PRO SSD with Heatsink 2TB PCIe Gen 4 NVMe M.2 Internal Solid State Hard
    Asset Drive: Same
    Power Supply: ASUS Rog Thor 1200 Certified 1200W 
    Operating System: Windows 10 Version 22H2 build 19045.2130
    Nvidia Drivers Version: 522.25
    Daz Studio Version: 4.21
     

    GPU Benchmark Results
    DAZ_STATS
    2022-10-26 14:20:22.413 [INFO] :: Total Rendering Time: 1 minutes 21.71 seconds
    IRAY_STATS
    2022-10-26 14:21:30.517 Iray [INFO] - IRAY:RENDER ::   1.0   IRAY   rend info : CUDA device 0 (NVIDIA GeForce RTX 4090): 1800 iterations, 0.957s init, 79.707s render

    Iteration Rate: (22.029) iterations per second
    Loading Time: (2.003) seconds

     

    I think I calculated the loading time correct. Just got my new system built out and wanted to test the RTX 4090. Here are the results. 

  • Thought I'd give this a try.

    System Configuration
    System/Motherboard: BRAND MODEL
    CPU:  Intel i7-11700kf @ 3.6Ghz
    GPU: NVidia RTX3060
    System Memory: Corsair 64GD (4x 16GB) 3200 
    OS Drive: 2TB Seagate HDD
    Asset Drive: same as OS
    Power Supply: 850W
    Operating System: Win 11 Home 64 bit 10.0 build 22621
    Nvidia Drivers Version: 31.0.15.1748
    Daz Studio Version: VERSION BITS
    Optix Prime Acceleration: STATE (Daz Studio 4.12.1.086 or earlier only)

    Benchmark Results

    GPU Only

    2022-10-26 15:37:42.394 [INFO] :: Total Rendering Time: 5 minutes 30.15 seconds

    IRAY:RENDER ::   1.0   IRAY   rend info : CUDA device 0 (NVIDIA GeForce RTX 3060): 1800 iterations, 8.014s init, 317.868s render


    GPU/CPU

    2022-10-26 15:43:30.624 [INFO] :: Total Rendering Time: 5 minutes 4.6 seconds

    IRAY:RENDER ::   1.0   IRAY   rend info : CUDA device 0 (NVIDIA GeForce RTX 3060): 1619 iterations, 1.152s init, 301.211s render

  • ScynothScynoth Posts: 2
    edited October 2022

    Didn't mean to but I did a duplicate post, not sure how to delete. 

    Post edited by Scynoth on
  • hamadahmadhamadahmad Posts: 12
    edited October 2022

    JB007 said:

    Is it fair to say then that the 4090 is working in Daz ... just not with all the taps turned on yet? Faster than the 3090 in Iray .. but not just by much yet?

    I've got my 4090 system ready to be built .. but I don't want to go for it just yet and be sitting around for 4 to 6 weeks (potentially without dForce) with something that's just a little bit faster than what I have with my 3090 right now .. 

    So is the 4090 cooked or should I wait until the oven is turned up first?

    From what I've observed. The answer to your question is Yes. Yes, it is working but only a 2-3 minute difference between the same render done with 3090 and 4090. And also deforce is not working with 4090. 

    So, I would say if you have a 3090 or 3090 ti there's no need to jump to 4090 at the current moment.

     

    Post edited by hamadahmad on
  • flyinglemmingsflyinglemmings Posts: 3
    edited October 2022

    System Configuration
    System/Motherboard: MAG B650M MORTAR WIFI (MS-7D76)
    CPU: AMD Ryzen 9 7950X @ 4.5 GHz(default)
    GPU: COLORFUL GeForce RTX 4090 24GB BattleAx Deluxe @ 2535 MHz (default)
    System Memory: Kingston DDR5-5200 32GB*2 @ 5200
    OS Drive: WD_BLACK SN770 2TB
    Asset Drive: Same
    Power Supply: SUPER FLOWER LEADEX HG 850W
    Operating System: Microsoft Windows 11 (10.0) Professional 64-bit   (Build 22000)
    Nvidia Drivers Version: 522.30
    Daz Studio Version: 4.21
     

    Benchmark Results
    DAZ_STATS
    2022-10-28 15:58:58.278 [INFO] :: Total Rendering Time: 1 minutes 22.85 seconds

    IRAY_STATS
    2022-10-28 15:59:17.025 Iray [INFO] - IRAY:RENDER ::   1.0   IRAY   rend info : CUDA device 0 (NVIDIA GeForce RTX 4090): 1800 iterations, 1.270s init, 80.120s render

     

    Post edited by flyinglemmings on
  • KeikuKeiku Posts: 143
    edited October 2022

    hamadahmad said:

    JB007 said:

    Is it fair to say then that the 4090 is working in Daz ... just not with all the taps turned on yet? Faster than the 3090 in Iray .. but not just by much yet?

    I've got my 4090 system ready to be built .. but I don't want to go for it just yet and be sitting around for 4 to 6 weeks (potentially without dForce) with something that's just a little bit faster than what I have with my 3090 right now .. 

    So is the 4090 cooked or should I wait until the oven is turned up first?

    From what I've observed. The answer to your question is Yes. Yes, it is working but only a 2-3 minute difference between the same render done with 3090 and 4090. And also deforce is not working with 4090. 

    So, I would say if you have a 3090 or 3090 ti there's no need to jump to 4090 at the current moment.

     

    Some other posts have mentioned it, but dForce is not working because of the NVIDIA 522.25 and/or 522.30 driver (rolling back fixes it). Unfortunately though, I belive the 4090 needs that one. However, from what I understand, they are working on a fix.

    Post edited by Keiku on
  • KeikuKeiku Posts: 143

    As a quick aside. I see some of you using the new AM5 platform with the 7950x. Any benefits seen there for Daz?

  • Keiku said:

    As a quick aside. I see some of you using the new AM5 platform with the 7950x. Any benefits seen there for Daz?

     

    All software uses thread(s) to complete tasks. In Daz3D, there are a lot of the bottlenecks that are due to single threaded actions/jobs (ex of one such bottleneck: file reading of morphs when a character loads).

    I suggest you look at this: https://www.cpubenchmark.net/singleThread.html

     

     

  • chrislbchrislb Posts: 100

    Keiku said:

    hamadahmad said:

    JB007 said:

    Is it fair to say then that the 4090 is working in Daz ... just not with all the taps turned on yet? Faster than the 3090 in Iray .. but not just by much yet?

    I've got my 4090 system ready to be built .. but I don't want to go for it just yet and be sitting around for 4 to 6 weeks (potentially without dForce) with something that's just a little bit faster than what I have with my 3090 right now .. 

    So is the 4090 cooked or should I wait until the oven is turned up first?

    From what I've observed. The answer to your question is Yes. Yes, it is working but only a 2-3 minute difference between the same render done with 3090 and 4090. And also deforce is not working with 4090. 

    So, I would say if you have a 3090 or 3090 ti there's no need to jump to 4090 at the current moment.

     

    Some other posts have mentioned it, but dForce is not working because of the NVIDIA 522.25 and/or 522.30 driver (rolling back fixes it). Unfortunately though, I belive the 4090 needs that one. However, from what I understand, they are working on a fix.

    You can also download an OpenCL driver for your CPU and have the CPU do dForce instead of the GPU if you have a 4090 or can't roll back the driver for some other reason. 

  • Aiight...so I'm having a tough decision here.

    I'm starved for VRAM at only 8GB on my 3070 right now.  This makes larger scenes unbearably slow when it falls back to a CPU render.

    And looking at the results people are getting with the 4090, along with the grim possibility of it melting the adapter with a few weeks of use...  Makes it seem the 3090 would be the way to go?  It may be a while before Daz fully utilizes the cores that the 40-series has, and nVidia may wind up completely redoing the power connector for it.  They're also $1000CAD cheaper than the 4090.

  • junkjunk Posts: 1,347
    edited November 2022

    DUAL RTX 3090 overclocked to +1000 MHz VRAM, +167 MHz GPU.  Both are hybrid water cooled.  In a hurry to get house work done so not fully following the guidelines.  I'll clean it up later but just wanted to try and get under one minute with two 3090's.  Of course I exited Daz Studio each time, killed it in the task manager, loaded the scene in "Texture Shaded" not iray, didn't touch iterations, used photorealistic mode.... yada yada.  This is probably not stable and I will most likely back it off to +800 MHz on the VRAM and +140 MHz on the GPU.

    Power draw at the wall fluctuating from ~600 to 940 watts...  ouch.

    Using Daz 4.21.05
    NVidia driver 517.40
    CPU: Ryzen 5600x overclocked to 4.7GHz
    Memory: 64GB DDR-3400
    Windows 10 Pro

    GPU1: EVGA Kingpin RTX 3090 Hybrid
    GPU2: EVGA GeForce RTX 3090 XC3 ULTRA HYBRID GAMING

     

    FROM THE LOG FILE:
    2022-11-06 10:22:04.180 [INFO] :: Total Rendering Time: 59.83 seconds
    2022-11-06 10:22:06.614 Iray [INFO] - IRAY:RENDER ::   1.0   IRAY   rend info : Device statistics:
    2022-11-06 10:22:06.614 Iray [INFO] - IRAY:RENDER ::   1.0   IRAY   rend info : CUDA device 0 (NVIDIA GeForce RTX 3090): 891 iterations, 1.299s init, 56.758s render
    2022-11-06 10:22:06.614 Iray [INFO] - IRAY:RENDER ::   1.0   IRAY   rend info : CUDA device 1 (NVIDIA GeForce RTX 3090): 909 iterations, 1.243s init, 56.286s render

     

    Post edited by junk on
  • outrider42outrider42 Posts: 3,679

    The Iray Dev Team just announced that a beta SDK of Iray 2022.1.0 is released. Of course, this does not mean Daz Studio will have it right away, only that it is available to Daz to start implementing it. There is also the question of if Daz will use a beta version of Iray, I don't remember. I believe they will use them, at least in the Daz Studio Beta branch.

    Here are some highlights in the notes:

    Iray 2022.1.0 beta released

    Feature Highlights:

    Guided sampling during camera movement

    Guided sampling support for the caustic sampler

    Much faster material and environment updates

    New spectral sun&sky model

    Native Lovelace & Hopper support

    Native Sphere/Particles support

    MotionBlur support for fiber/curve vertex motion

    Catmull-Rom and linear data support for fibers/curves

    ..and the usual pile of bugfixes

    As you can see, this update adds native support for both Lovelace and Hopper GPUs, so perhaps that will unlock its performance. They made no mention of performance.

    The 4090 vs the 3090 can be a tough choice if the 3090 is on a good deal. You can get two 3090s and even combine their VRAM with Nvlink (at least for texture data, mesh data is not shared). So it is possible to get more VRAM if it really is that important (or of course, get the A6000.) The 4090 has no Nvlink. So the 3090 can still offer an advantage over the 4090 if you choose to pair them. Also, two 3090s are currently faster than a single 4090. At least for now. We don't know how much native support will boost performance, or if it will boost it at all for that matter. I assume it will, because otherwise it would feel lackluster when most other render engines are getting double or more performance with a 4090 over a 3090/ti.

  • JB007JB007 Posts: 118

    From the Iray Dev Blog (https://blog.irayrender.com/post/700645909939339265/so-far-testing-iray-with-the-4090-is-not-yielding)

    Anonymous asked:

    • So far testing Iray with the 4090 is not yielding the expected performance enhancements when compared to other rendering engines (versus previous flagship 3090ti). Other engines tend to double (or more) performance with a 4090, while Iray is maybe 70% faster at best, sometimes just 60% faster. Is this as good as the 4090 can get with Iray? Or can an updated Iray with the latest CUDA support improve performance? On that note, when can we expect the next version of Iray SDK?
      • The next Iray SDK (2022.1.0 final) is due beginning of december. Some Ada/Lovelace performance improvements may still be possible for that one, but from our own tests we already actually see an average ~2x improvement comparing various scenes on a 4090 vs a 3090. Will post benchmark results soonish, mileage varies heavily on scene types.
  •  

    Very interesting results. 
    I uninstalled 4.21.0.5 in DAZ Install Manager (64-bit). After installed 4.16.0.3

    4.21 Total Rendering Time: 12 minutes 5.46 seconds
    4.16 Total Rendering Time: 9 minutes 7.73 seconds
    3min???

    System Configuration
    System/Motherboard:  Asus Z170-A 
    CPU: Genuine Intel(R) CPU 0000 @ 2.80GHz (Engineering Sample) QTJ0 4,5GHz analog i9 9900k
    GPU: NVIDIA GeForce GTX 1070 Ti  (8 Gb)
    System Memory: 4 x 8GB DDR4-2666
    OS Drive:  SSD 512Gb KingSpec (NE-512)
    Asset Drive:  WDC WD5003AZEX-00MK2A0  (500 Gb, 7200 RPM, SATA-III)
    Power Supply: Chieftec SLC-650C 650W
    Operating System: Windows 10 Pro
    Nvidia Drivers Version: 522.25
    Daz Studio Version: 4.21.0.5 vs 4.16.0.3

     

    2022-11-11 23:04:13.927 [INFO] :: Total Rendering Time: 12 minutes 5.46 seconds

    2022-11-11 23:04:17.505 Iray [INFO] - IRAY:RENDER ::   1.0   IRAY   rend info : CUDA device 0 (NVIDIA GeForce GTX 1070 Ti): 1800 iterations, 1.349s init, 721.628s render

    2022-11-11 23:09:21.976 --------------- DAZ Studio 4.21.0.5 exited ------------------
     

    2022-11-12 01:40:42.346 +++++++++++++++ DAZ Studio 4.16.0.3 starting +++++++++++++++++

    2022-11-12 01:51:27.382 Total Rendering Time: 9 minutes 7.73 seconds

    2022-11-12 01:54:25.800 Iray [INFO] - IRAY:RENDER ::   1.0   IRAY   rend info : CUDA device 0 (NVIDIA GeForce GTX 1070 Ti):                  1800 iterations, 2.060s init, 543.199s render

     

  • I want to be sure I read the numbers in that table right.  Vanilla RTX 2070 has 4.531 iterations per second, and the 3060 Ti gets 10.648 iterations per second?  There's no 3060 12GB, which isn't the Ti version, or vanilla 3060 but I'm guessing it'll be within a few percent of the Ti version.  Does that sound right to you?  It's a huge boost compared to the 2070 which is why I doubted it.

  • outrider42outrider42 Posts: 3,679

    Pickle Renderer said:

    I want to be sure I read the numbers in that table right.  Vanilla RTX 2070 has 4.531 iterations per second, and the 3060 Ti gets 10.648 iterations per second?  There's no 3060 12GB, which isn't the Ti version, or vanilla 3060 but I'm guessing it'll be within a few percent of the Ti version.  Does that sound right to you?  It's a huge boost compared to the 2070 which is why I doubted it.

    Ampere is just that good. The ray tracing combined with how CUDA is split (note the CUDA counts) make Ampere shine for Iray and other rendering engines. To be fair, you cannot directly compare results from different versions of Iray and Daz, and the 2070 result is on a much older version. Another factor is the impact of ray tracing will vary on different scenes. So some scenes may see a larger gap, and some may see a smaller gap. Still, the results can give a general idea.

    But yes, the 3060ti really is a lot faster for Iray. It should easily beat the previous gen flagship 2080ti. The 3060 is actually cut down a fair bit compared to the 3060ti, so don't expect quite that level. However the 3060 is close to the 2080ti if I recall right. That just shows how fast Ampere is for Iray.  The 2000 series was a huge jump over the 1000, and the 3000 was yet another huge leap. So pretty much every 3000 card is as fast or faster than the 2080ti, which is pretty wild. Only the lowest tier 3050 is slower than the 2080ti. The 3060 for its part brings 12gb to the table, making it very unique in the lineup.

    We may not see these kinds of leaps again.

  • outrider42 said:

    The 3060 for its part brings 12gb to the table, making it very unique in the lineup.

    That's why I was looking at it.  I continue to be shocked by card prices, looking at the 4090 and published prices of the 4080 (3080, 3090 are still ridiculous).  NVIDIA hasn't given the 4070 price yet but I expect it'll come in around £850.  And that's also a 12GB card.   But I can get a 3060 for £389.  So those are the choices.  Every £ I spend on a card is a £ not in my savings account... I'm starting to think the value proposition here is the 12GB 3060.  I'll get quite a boost from it for productivity work but not for gaming, though I'm thinking perhaps the RT pipeline in games like Cyberpunk will be useable compared to the 2070...

     

    Hmmm.

  • outrider42outrider42 Posts: 3,679

    Its tough because the 3060 isn't much better than the 2070 at other things. There may be some scenes where the 3060 doesn't stretch out over the 2070 as much, again it depends on what you put in the scene.

    Another option is to use both cards at the same time, though the 2070 will drop out if the scene exceeds its 8gb. This also uses more power and space, of course.

    Nvidia allegedly is very intent on keeping prices high, and by not releasing any card that competes with the 3090 directly (like the "4070" would have), that allows prices to stay high. Perhaps AMD can force the issue since their top card is an even $1000. We can't use AMD cards with Iray, but we can still benefit from good competition.

  • I think I'll pull the trigger on a 3060 12GB then, and wait another 12 months or so to see what the 4xxx series lineup ends up looking like. 

  • outrider42outrider42 Posts: 3,679

    You could maybe wait just for the 4080 to release since it is just days away. I'm not saying to buy one, but rather its release might help just a little bit. New cards have to compete with used cards on eBay, too. I don't think the 4080 will be received very well, but it could still shift prices a little by simply being there. Retailers probably have to make room for these, so that means something has to go. There could be some better sales on the 3060 as a result.

Sign In or Register to comment.