Differences
This shows you the differences between two versions of the page.
Previous revision | |||
— | appwiki:octane [2023/01/05 08:49] (current) – [Octane Render info] ying | ||
---|---|---|---|
Line 1: | Line 1: | ||
+ | ====== Octane Render info ====== | ||
+ | |||
+ | (updated: 2023.01) | ||
+ | |||
+ | * free blender-octane-edition download: (also available for Unreal+DAZ) | ||
+ | - reg acc: https:// | ||
+ | - install BlenderOctaneEdition (octane render prime for BL) + OctaneServerPrime: | ||
+ | - run OctaneServerPrime, | ||
+ | - it may ask you to download and put cudnn library, just click its Download button, it will do everything for you. | ||
+ | - once done | ||
+ | - click Active button, and login in pop windown, then it shows status: activated for 1 month (need stay online) | ||
+ | - start OctaneBlenderEdition, | ||
+ | - activate Octane plugin in Edit > Preference > Addon | ||
+ | - also, Window > Octane DB to load all the material presets | ||
+ | - basic initial setup | ||
+ | - Render Property Tab: engine = Octane | ||
+ | - select main cam, Camera tab: check " | ||
+ | - Render Property Tab: Color Management > View Transform: RAW, gamma = 1 | ||
+ | - Environment tab: Octane Environment = daylight environment | ||
+ | * ref: | ||
+ | * https:// | ||
+ | ====== Octane GPU Render Hardware ====== | ||
+ | |||
+ | * Octane render speed (version 3 and 4) directly related to GPU power and count | ||
+ | * octane 3.07 and below don't support RTX 20xx and GTX 16xx GPU, | ||
+ | * octane 3.08 and 4 above support RTX 20xx and GTX 16xx GPU, | ||
+ | * octane 3 and octane 4 use different benchmark points | ||
+ | |||
+ | ===== RAW GPU Power Compare ===== | ||
+ | |||
+ | 2022 update | ||
+ | | gpu | cost2022 | octane bench 2020 | | ||
+ | ^ 4090 | 2300 ^ 1275 | | ||
+ | | 4070 ti | | 711 | | ||
+ | | 3090 ti | | 689 | | ||
+ | | 3080 ti | | 643 | | ||
+ | | 3080 | | 557 | | ||
+ | ^ 3070ti | ||
+ | | 3060 ti | | 366 | | ||
+ | | 3070ti mob | | 352 | | ||
+ | | 2080ti | | 344 | | ||
+ | | 2070 sup | | 256 | | ||
+ | | titan xp | | 214 | | ||
+ | | 1080ti | ||
+ | ^ 1070ti, | ||
+ | | 1650 sup | | 93 | | ||
+ | ^ 1060 | ^ 75<< | | ||
+ | |||
+ | | GPU | cost2019 | Octane bench 3.06 (eff) | octane 4 | Watt | Power Pin count | PSU | total Eff | | ||
+ | | 1060 3GB | 280 | 79 (0.28 ppd) | 80 (0.29) | 120W | 6pin | S12II 520W (75) | 0.22 | | ||
+ | | 1070 Ti | 600 | 135 (0.22 ppd) | 153 (0.26) | ||
+ | | 1070 Ti + 1060 3GB | 880 | 228 (0.26 ppd) | 240 (0.27) | 300W | 6+8pin | S12II 620W (91) ^ 0.23 | | ||
+ | | 2x 1070 Ti | 1200 ^ >> 273 (0.23ppd) << | ||
+ | | 3x 1070 Ti | 1800 | >> 386 (0.21ppd) << | ||
+ | | 4x 1070 Ti | 2400 | 507 (0.21ppd) | ||
+ | ^^ | ||
+ | | 1070 Ti 2nd hand | 380 | 135 (0.36ppd) | ||
+ | | 1080 Ti 2nd hand | 700 | 185 (0.26ppd) | ||
+ | | 1080 Ti | 1200 | 185 (0.15ppd) | ||
+ | | 2x 1080 Ti | 2400 | 375 (0.16ppd) | ||
+ | | 3x 1080 Ti | 3600 | 553 (0.15ppd) | ||
+ | ^^ | ||
+ | | RTX 2070 | 710 | | 209 (0.29) | 175W | 8pin | S12II 520W (75) | 0.27 | | ||
+ | | RTX 2080 | 1100 | ||
+ | | RTX 2080 Ti | 1725 | | 303 (0.18) | 250W | 8+8 | S12II 520W (75) | 0.17 | | ||
+ | ^^ | ||
+ | | 2x RTX 2070 | 1420 | | 409 (0.29) | 350W | 2x8 | S12II 620W (91) | 0.27 | | ||
+ | | 3x RTX 2070 | 2130 | | >>621 (0.29)<< | ||
+ | | 2x RTX 2080 | 2200 | | 441 (0.2) | 430W | 2x8+2x6 | PrimeUltra1000GD (240) | 0.18 | | ||
+ | | 3x RTX 2080 | 3300 | | 582 (0.18) | 645W | 3x8+3x6 | Prime 1300GD (293) | 0.16 | | ||
+ | ^ 2x RTX 2080 Ti | 3450 | | 588 (0.17) | 500W | 2x8+2x8 | PrimeUltra1000GD (240) | 0.16 | | ||
+ | | 3x RTX 2080 Ti | 5175 | | 849 (0.16) | 750W | 3x8+3x8 | Prime 1300GD (293) | 0.16 | | ||
+ | |||
+ | * Conclusion: | ||
+ | * (3300 SGD budget - 588 @O4) for RTX 2080 or 2080 Ti? | ||
+ | * Go for **2x RTX 2080 Ti**, it is better than 3x RTX 2080 setup, 4 PCI-E connection better to manage. | ||
+ | * (2200 SGD budget **621** @04) for 2x RTX 2080 or 3x RTX 2070? | ||
+ | * definitely for **3x RTX 2070**, only 3 PCI-E connection to manage and 50% more power. | ||
+ | * (700 SGD budget **209** @O4) RTX 2070 vs GTX 1080 Ti 2nd hand? | ||
+ | * for octane 4, definitely go for RTX 2070, it is better and newer | ||
+ | * for octane 3, you can only go for 1080 Ti | ||
+ | * (700 SGD budget **312** @04 2nd hand) RTX 2070 vs GTX 1070 Ti 2nd hand? | ||
+ | * for price of RTX 2070, you can get **2x GTX 1070 Ti 2nd price**. | ||
+ | |||
+ | **Final Conclusion** | ||
+ | * 2070 and 1070 single 8pin PCI-E power connection is great for multi-GPU and easy Power Supply setup. as even 850W PSU can power 3 of them. | ||
+ | * for 2nd hand price, it is better to buy 3x GTX 1070 Ti at 380x3 = 1140 SGD with 850W PUS (140SGD). | ||
+ | * total = 446 @ O4 score, best for price, only outperformed by 3x GTX1080 Ti (more expensive) | ||
+ | * for brand new card, it is better to buy 3x RTX 2070 at 710x3 = 2130 SGD with 850W PUS (140SGD). | ||
+ | * total = 621 @ O4 score, best for price, only outperformed by 3x 2080 Ti (more than double price.) | ||
+ | |||
+ | ===== Power Supply Connection Consideration ===== | ||
+ | |||
+ | | PSU (Seasonic) | ||
+ | | gm750W | ||
+ | | s12iii 650w | | 93 | | ||
+ | | PSU (Seasonic) | ||
+ | | S12II 520W | 2 PCI-E max | 75 | | ||
+ | | S12II 620W | 2 PCI-E max | 91 | | ||
+ | | M12II 850W | 3 PCI-E max | 140 | | ||
+ | | PrimeUltra1000GD | 4 PCI-E | 240 | | ||
+ | | Prime 1300GD | 6 PCI-E | 293 | ||
+ | |||
+ | ===== Motherboard PCI-E x16 Port Count ===== | ||
+ | |||
+ | - **for mATX motherboard, | ||
+ | * and 2nd slot may block front panel pin connection, so you can use front usb | ||
+ | * example: Asrock B350m Pro4 | ||
+ | - **for ATX motherboard, | ||
+ | * example: Asrock B350 Pro4, B450 Pro4 | ||
+ | - **for better ATX motherboard, | ||
+ | * however, some motherboard pin may be blocked by last GPU, but some motherboard has better pin arrangement for this issue | ||
+ | * example: | ||
+ | * Asrock X370 Taichi | ||
+ | * MSI X370/X470 GAMING PLUS AM4 | ||
+ | - **For 4x GPU or better 3x GPU setup**, only high-end CPU board, like X399 series board for AMD TR 2950X can have 4 slot | ||
+ | * some cheap x399 board has 4 slot, but spacing is better which cause only 3 slot can take GPU, like MSI MEG X399 CREATION sTR4 | ||
+ | * example: | ||
+ | * ASRock X399 Taichi sTR4, Asrock Fatal1ty X399, 4 Slot with 2 card spacing. | ||
+ | * GIGABYTE X399 AORUS Gaming 7 sTR4 has 5 slot, but only 4 slot can be used to take 2 card spacing | ||
+ | |||
+ | |||
+ | |||
+ | **Additional Concern** | ||
+ | * some riser card may solve the block pin issue, but not ideal for both case and stability | ||
+ | * http:// | ||
+ | |||
+ | ===== GPU fan type Blower vs Fan and GPU Size ===== | ||
+ | |||
+ | * for 3 GPU setup or better 2 GPU, those motherboards are better spaced for either fan design type GPU, but of course, Trip fan type are better than blower type in this case | ||
+ | * for 4 GPU setup or those 4 slot motherboard, | ||
+ | * Better Casing cooling and Case fan flow control (cleaning and maintain dusts) are always better to keep GPU at high load. | ||
+ | |||
+ | * Those multiple GPU oriented maker all make blower type, like leadtek, zotac, and of course expensive Asus. | ||
+ | |||
+ | * Compare Blower and Fan type extreme case: | ||
+ | * https:// | ||
+ | |||
+ | |||
+ | **GPU if too long, it can also block SATA connectors and USB3 contector in some case** | ||
+ | * like leadtek GTX 1070 Ti WindForce is too long in some case and motherboard combination | ||
+ | * some small card like Zotac 1070 Ti mini is small like normal 1060 cards, can fit better in either main slot or last GPU slot without causing pin blocking. | ||
+ | ===== Reference ===== | ||
+ | |||
+ | * Octane Benchmark | ||
+ | * octane 3.08, RTX vs GTX: https:// | ||
+ | * Octanebench 4 for RTX and GTX: | ||
+ | * https:// | ||
+ | * Octanebech 3.06 for GTX: | ||
+ | * https:// | ||
+ | * Mix GPU render result: | ||
+ | * https:// | ||
+ | * RTX GPU vs GTX GPU in 3.08 | ||
+ | * https:// | ||
+ | |||
+ | * Calculation | ||
+ | * Power calculator: http:// | ||
+ | * Price to Power table: https:// | ||
+ | |||
+ | * Build Guide | ||
+ | * https:// | ||
+ | * https:// | ||
+ | * https:// | ||
+ | * https:// | ||