@Grazie@BruceUSA Grazie, although we are all interested in how your new system performs, don't get hung-up on one 14s test... BruceUSA & I have built our own rigs & know how to fix the settings if our overclocks fail...The key is comparing your new system to your old system and I think you will be VERY pleased even if you don't get to 14s on ONE test...
My 9900K w/Vega 64 (currently) completes the Red Car Test in 14s average compared to my Xeon w/RX 570's score of 35s, so about 2.5X as fast on this ONE test. However, I ran 10 different 1 minute sample projects of my own work on both systems and found that the 9900K ranges anywhere from 1.75X to 2.5X as fast depending on the source video, FX, render codecs, etc. (2.5X was using the same Magix HEVC codec used for the Red Car Test - so the Magix codecs do appear to tap into the strengths of newer CPUs/GPUs...) IMO your new rig may not reach 14s on the Red Car Test, but be faster in 9/10 other tests...
This is why I wanted BruceUSA to perform: ParrallaZ Rock 01 Test to MAGIX AVC/AAC MP4 Internet HD 59.94 fps.
I have responded to your previous request about the test. Its slow and I did not care to finish it and abort and delete the file. But if you want to compare your to mine in regards to performance in Vegas . Why don't you try 2-3 tracks in multicam editing with gh5 10 bit 4k footage and vegas project set to 32 bit and to best/full. Add 3 fx on it also. See if you can muster it in real time multicam editing. I want to seea video capture prove that you can. I have my video posted on Vimeo already and my system able muster 2 tracking on those settings I mention above. Also I want to see can you render 1080P and get 160 frames that is 5x in realtime. Can your mighty 9900k can cut it. Prove it. I already had many sample videos and screen shot posted on this forum. You can find it if you want.
@TheRhino - Hi there! Yes I was wanted to get a Real World datum comparing my existing Box to that of the new Box and attempt to ascertain the speed value and extent of my investment. But more than that, it is to determine how “quick n slick” I can get to my narrative and how organically I can achieve it. I’m hoping for 5x that which I’m currently realising plus the fluid and responsiveness of several plugs I’m operating.
The 14s Club is most certainly a valuable target to check or compare against. I’m in awe at the compounded knowledge you all have gleaned over the years.
Sorry I didn't come across this thread till now because I finished a new build myself a couple weeks ago and wrestled with allot of the same issues as Grazzie... well when you build it yourself it's not ever really finished. I kind of documented the assembly on a facebook thread if anyone is interested. It's essentially an ASUS mb, 9900k cpu, 32 gb 3600 mhz g.skill memory, and Radeon 7. All air cooled. Been beating it to death testing and knocking off a few 1-hour projects.
Should mention that I'm doing strictly HD 1080 at this point so my considerations and experience might not be applicable to those doing 4k. (but I'll eventually get into that I'm sure) Fwiw, my air-cooled stock setup could not make it though 5 minutes of p95 without throwing errors. Though it behaved fine on Red Car and rendering a typical 57-minute, 3-camera catv show I do. Brought the memory timing down to 3400 mhz and lowered maximum turbo boost from 5ghz down to 4.8 ghz and it ran p95 for days without error. But I noticed 2 of the 8 cores occasionally hit 100C on 4k ffts. But none ever past the 80's in Vegas. I'm good with that for now but I've gotten my hands on an interesting hybrid cooler that looks most promising... it's a Phononic HEX 2.0 thermoelectric. Curious if anyone else here has used it. Haven't popped mine in yet.
My Red Car has pretty consistently been 13 to 14 seconds. But for me, the rendering performance of one of my 57-minute 3-camera HD projects is more significant because its typical of what I do the most of. Curiously, when I brought my memory and turbo clock speed down there was no change in my rendering benchmark times. Same putting my Red Car and typical show onto pcie m.2 vrs the el cheapo 512gb sata ssd's I use for backing up and archiving my projects (Pioneer and Silicon Power). Same for knocking the pcie Radeon slot down to pcie2 and/or x8. Although there were noticeable differences in project load and media copying times to sata ssds. Again, all of this might not apply to 4k.
@Howard-Vigorita - Firstly, thank you! - That must have taken you time to write-up but all that testing? Secondly, I understood ALL the tech stuff with the exception of the deeper significance.
As you're on a similar wavelength as I when getting to Real-World outcomes, here's my brief OLD and NEW Excel ScreenGrab:
And yes, presently 1080p, s l o w move over to 4k.
So, Chums, I'm guessing I'll see a marked/better difference in Creation times - yes? I'll be interested to see if RAM Builds are able to be made BIGGER and are created faster.
Your thinking of getting aan RTX2070, right? You may want to opt for an RTX card with more memory however.
Why? Not for Vegas 15/16, but maybe for a future Vegas or other NLE, I upgraded recently Magix pro X to their new version 11 (I skipped last years extension), and I ran a few tests and the HEVC encoding is using NVENC very efficiantly and very fast, but what is more, when I used a HEVC 4K 50P source clip from my Gopro, it was using almost all of my 11GB of my video card memory (see screenshot), if MPX is any indication where Vegas is or can be going in the future with regard to HEVC HW acceleration, then your 8 GB on the RTX2070 is not going to make it...
Here's my latest updated test summary. My 3-year old Xeon system has 128 gigs of ram but I don't think Vegas ever used anywhere near that much... but I don't generally run more than 2 instances at once and only then if I'm copying text screens or credit rolls from one project to another. Also did a bunch of media testing, thinking maybe about a NAS or Windows remote file systems. Looks like a 2-nic team might be optimal but unfortunately Intel nics don't support that anymore with Win10... makes those affordable 10gig nics start lookin pretty good.
@Howard-Vigorita - I asked for Real-World, and you sure did that, and then some. Thank you. Revealing my ignorance, what’s a NIC? Also, would you care to add your own personal experience of the speed and efficacy of the actual editing experience for those three PCs?
Thanks for running these tests. It will help those planning new builds to see what components provide the best bang/buck... I lowered my 9900K CPU from 5.0 Ghz to 4.89 and ran the tests (w/ my slower DDR4 3000 memory). In NovaBench & Vegas, the cheaper Vega 64 compares very similarly to the Radeon VII... Also, unlike AMD's Threadripper which requires 4 sticks of fast DDR4, the 9900K performs very well with DDR4 3000...
Novabench RAM Speed = 30,211(6,136 slower than DDR4 3600 CPU Score = 1,788 @ 4.89 ghz vs. 1,808 @ 5.0 ghz Vega 64 Graphics = 1,173 (Vega64 same as VII) vs. 1,207 w/CPU @ 5.0 ghz Direct 3D11 fps = 203 (Vega64 same as VII) vs. 211 fps w/CPU @ 5.0 ghz CineBench ....OpenCL = 178.77 (VII is faster here...) ....CPU = 1,976 @ 4.89 ghz. vs. 2,126 @ 5.0 ghz Single Core = 208 @ 4.89ghz. vs. 213 @ 5.0 ghz Red Car Test Magix AVC 1080-30p (QSV) = 0:14 vs. 0:13 @ 5.0ghz... Magix AVC 1080-30p (cpu only) = 0:53 Magix AVC 1080-30p AMD/VCE = 0:34 (Vega64 same as the VII) Sony AVC 1080-30p (gpu) = 0:43 (Vega64 same as the VII) XDCamEX HQ (1080-30p) (gpu) = 0:13 (Vega64 same as the VII)
I am still looking forward to see some results from ALL the mighty 9900K to run some 2-3 tracks 4K 10 bit GH5 (specific) make your cut in multicam mode and in 32 bit Vegas project settings along with 3 FX (color correction, curve, sharpen) I say this because I constantly see this guy here mentioned and compared his mighty 9900K vs Threadripper. So, I already know what my threadripper can do in 10 bit 4K multicam in 32 bit with FX. But can yours? I ask for prove but so far I don't see a respond as obvious he can't cut it.
PS. 9900K is cheaper to build hell yes. 9900K performs better then TR, Hell NO, not even close in over all. I also edit photo in Lightroom CC. Export photo in Tiff or Jpeg file is an instantly done. The second I click export, BOOM done. No waiting, no lags, no nothing. I used to own a high clock intel 3930K 6 cores @ 5ghz and still own 4930K @ 4.5ghz both system has lags in export.
Here's a version of the Red Car with the video media trans-coded to a number of 4k formats. Haven't actually run any render timings with it myself yet but it seems to load and play smoothly and error free in Vegas versions I have handy... 13 and 16. The clip that's an mxf in the original is a 3 gb mov file in this one so it should stress both processor and storage media. Whole thing is about a 4 gb download.
Last year I was thinking about upgrading my i7-8700k CPU to an i9-9900k but figures I got from an i9-9900k user here in the forum showed me that it is not really worth the effort.
Here are my render times with the i7-8700k @ stock speed with an AMD RX570/8GB RAM
The i7-8700k is air cooled and the core temperature normally does not exceed 62°C. However, if I would build up a new system, I would take the i9-9900k or wait a little for the upcoming Intel CPU generation.
Former user
wrote on 6/17/2019, 5:38 AM
Here's a version of the Red Car with the video media trans-coded to a number of 4k formats. Haven't actually run any render timings with it myself yet but it seems to load and play smoothly and error free in Vegas versions I have handy... 13 and 16. The clip that's an mxf in the original is a 3 gb mov file in this one so it should stress both processor and storage media. Whole thing is about a 4 gb download.
The thing is @Howard-Vigorita, this was previously done, maybe a year ago roughly. So going forward, while I think its indeed a good idea, we would need to standardise on say for example your 4K version of the Red Car test . Maybe call it the HV4K Red Car test.
The original Red Car test has a pdf and its clear what templates to use for the avc and mxf renders.
For example, not dissing @BruceUSA but its only now became apparent (see his post today) that his 14s render time wasn’t avc but the xdcam render.
Just a suggestion, do a screen capture of ideally just one or at most two render templates that should be used with your HVRC test and include in a downloadable zip.
PS. 9900K is cheaper to build hell yes. 9900K performs better then TR, Hell NO, not even close in over all. I also edit photo in Lightroom CC.
The 1950x SHOULD perform better than the 9900K, but it does not in many apps because they have not optimized the apps for larger multi-core performance (yet)... According to Pugent Systems' benchmarks, the 9900K outperforms the 2950X in Lightroom (& Photoshop)
I am still looking forward to see some results from ALL the mighty 9900K to run some 2-3 tracks 4K 10 bit GH5 (specific) make your cut in multicam mode and in 32 bit Vegas project settings along with 3 FX (color correction, curve, sharpen)
I do not own a GH5 or currently have GH5 footage from clients to use as comparison... However, I understand that it is capable of 400 Mbit/s... My current project involves editing a 5 camera shoot using a clients' ProRes 422 HQ intermediates (4096x2160 @ 700+ Mbit/s).
The way I edit multicam in Vegas may vary from others, but my 9900K handles my workflow well... For 3-5 camera shoots I sync the clips by audio to the best (or edited) audio track & then resize each camera's separate video track into the 4 corners (& 5th in middle) so I can see all cams at once on the preview window's monitor. I then arm a mic and call-out the cam I want to switch to as I place a marker. Afterwards, a script makes all of the cuts at the markers & I drag or copy the selected cam to a new full-size track on top. Previewing 5 tracks at best/full quality is smooth with Vegas CPU usage at 52%, GPU at 18% & Disk at 225-250 MB/s. Same if Vegas is set to 32-bit floating point video levels... I typically do not have anything background rendering during this important step...
Here's a version of the Red Car with the video media trans-coded to a number of 4k formats. Haven't actually run any render timings with it myself yet but it seems to load and play smoothly and error free in Vegas versions I have handy... 13 and 16. The clip that's an mxf in the original is a 3 gb mov file in this one so it should stress both processor and storage media. Whole thing is about a 4 gb download.
The thing is @Howard-Vigorita, this was previously done, maybe a year ago roughly. So going forward, while I think its indeed a good idea, we would need to standardise on say for example your 4K version of the Red Car test . Maybe call it the HV4K Red Car test.
The original Red Car test has a pdf and its clear what templates to use for the avc and mxf renders.
For example, not dissing @BruceUSA but its only now became apparent (see his post today) that his 14s render time wasn’t avc but the xdcam render.
Just a suggestion, do a screen capture of ideally just one or at most two render templates that should be used with your HVRC test and include in a downloadable zip.
JN. No problem at all :) XDcam were always has been the fastest to render, even from the original test with PDF started by Hulk, I think. Don't remember the exact name. If I remember correctly, TheRihno here now is the same TheRihno back then. He participated in the test as well. I think at that time he has i7 980X with GTX580 or something. He inputs his results and left out the Blank Excel grid when we asks for more detail about the result. Then he disappeared. Because his claimed were kinda suspicious. At that time I was running a 5ghz 3930K 40 PCIe Lanes 6 cores i7.
Here's a snip from that original Sony pdf on the 2 render templates they used:
Finally, entire project renders were performed. In all cases, the encoders for both AVC and
XDCAM EX were used and a total time in seconds was recorded. For Vegas Pro 10, the
“Sony AVC Internet 1920x1080-30p” template was used. For Vegas Pro 11 the
“MainConcept AVC Internet HD 1080p” template was used. Both Vegas Pro versions
contain the same “XDCAM EX HQ 1920x1080-60i, 35 Mbps” template and that template
was used for testing both versions.
The only bone I have to pick with Sony is they compared AVC rendering 30p with XDCAM rendering 60i while knowing all the source clips were 30p. Even handicapped XDCAM EX HQ renders faster with better audio than AVC but the files are larger.
Here's a snip from that original Sony pdf on the 2 render templates they used:
Finally, entire project renders were performed. In all cases, the encoders for both AVC and
XDCAM EX were used and a total time in seconds was recorded. For Vegas Pro 10, the
“Sony AVC Internet 1920x1080-30p” template was used. For Vegas Pro 11 the
“MainConcept AVC Internet HD 1080p” template was used. Both Vegas Pro versions
contain the same “XDCAM EX HQ 1920x1080-60i, 35 Mbps” template and that template
was used for testing both versions.
The only bone I have to pick with Sony is they compared AVC rendering 30p with XDCAM rendering 60i while knowing all the source clips were 30p. Even handicapped XDCAM EX HQ renders faster with better audio than AVC but the files are larger.
You can change to 30P in XDCAM and that what I did for the above post here.
Former user
wrote on 6/17/2019, 10:34 AM
Probably best to start a dedicated benchmark thread, that way anyone that is looking for systems HW comparisons would at least have a one stop shop.