Greetings! I verbally provided the CL in the intro for XMP but the DDR5 is 40-40-40 1.25V and the DDR4 is 19-25-25-45 1.45V ALSO..! If you get a blank white page on the InvMark site, there's been a BUG since day 1 with browser tracker blockers. If you have a tracker blocker i.e. Bitdefender it will cause the site to fail to load. Cadac are looking into it but its extremely difficult to diagnose. If you want to use InvMark just disable the tracker blocker for that site. For the benefit of anyone new, the InvMark test includes a HUGE dataset of around 1000 unique files and the final score includes a grading of how the system performed whilst opening that dataset from disk on several different occasions, there's also a save to disk test, a data translation test, so InvMark isn't just a straight up plain simple "cad model a part" test by any stretch of the imagination! Also, my day 1 AutoCAD CPU index was also bang on 488 with the 12900K so I honestly don't know why the Revit results changed so dramatically, it definitely can't be thermals as that wouldn't be affected so I don't know. Still, interesting none-the-less.
Small observation made in Fusion 360 that might not have anything to do with it. NVidia studio drivers on a RTX3060 (laptop) kept it at 900MHz and even put it at 210MHz very frequently, even when trying to rotate and zoom the part. The Game Ready drivers kept it at 1425 MHz (with pushing it upwards, same as the studio drivers) and went to idle a lot less. So, maybe the Quadro driver caused the difference by keeping the card running slightly slower?
@@ExMachinaEngineering Something to investigate possibly, would be pretty bad for PR if the Quadro line was causing that much of a slow down. I'm all for longevity and reliability in the professional products but not if it means that much of a performance hit under an entry level GeForce card
Thanks for spreading useful information for professionals. There are tons of gaming tests for hardware but only few for professional application. I appreciate the inclusion of Revit even though it's not your domain. Tom
Exactly what I needed to know for a necessary purchase for employment. Saves a decent amount of the painfully expensive Revit PC. It’s the equivalent of buying an HGV instead of a car. Not exotic but definitely expensive because of what it needs to be capable of
I always support these kind of work since everyone's testing is pointed at gaming like there is nothing else in this world except games...is it possible that nobody ask "Who's making them?"
Where I live DDR 5 Ram is currently double the price of DDR 4. Performance on both gaming and Autodesk products are the same between the 2 RAM generations. Disappointing to say the least.
I'm doing "as-built" models in Revit, which is modeling off of a pointcloud. Pointclouds can be anywhere from 6GB to 300GB. I was maxing my 32GB ram kit and getting lots of stuttering. Now on 64 and smooth as butter. If I'd build my system on a ddr5 platform (MOBO+64 GB RAM), I'd have to spend about $800 on these, while my DDR 4 (b550+4X16) was about $450. The only downside to my system is that I won't be able to have more RAM if I'll need it in the future (I have 4X16 now). So if I was building a system today, I would have gone with the DDR 5. But for most users, DDR 4 will be enough for YEARS to come.
Thanks for putting this together. It's just what I was looking for today. Have you benchmarked the integrated graphics? I don't do much 3d work in AutoCAD or Revit and wonder how much a quadro or gaming card helps 2d performance, particularly compared to Intel's latest UHD Graphics 770.
Its on the to-do list if my Z690 board is still working, there's a chance it might have been cooked by a faulty part that I was testing but if not then yep I'll run some tests on that
I don't know if this helps but i have a laptop with an AMD apu. My work is mainly 2D CAD and i found that the apu is holding good, except that it needs a larger ram than 16GB in order for a headache-free hatching on Autocad.
I'm building a new system, 3080 ti, 12700k, dont want to spend more on ram so i will use my old 64gb of ddr4 2993mhz cl21, will it make a difference if I use that or should i get faster 3200-3600 cl16 or lower ram? I plan on running solidworks
I think ddr3 would be the same story as well. Other then PCIe 4.0 NvME support, nothing has incredibly upgraded on PC side. I mean, i9 12900k is only %50 faster then my i7 4770k (stock)? What?
boring huh ? Because... it was obvious :) The thing here is, these applications simply DO NOT need and CAN NOT UTILIZE 90GB/s bandwidth available from DDR5 memory modules. Bugatti Veyron won''t be faster than Opel Astra on a road with radars/cameras where 40mph is the limit. Both will be doing 40, that's it. Anyone running DDR4 at 3200 or 3600MHz is absolutely fine. You didn't state timings, primary and secondary, but as we see it doesn't make any difference. Uh, obviously :) Day1 performance delta : that's not so obvious :) we guess you didn't notice any temperatures deltas on the display of liquid cooler, otherwise you would state that in video.
I wasnt monitoring the temperatures because it wasnt a day 1 vs day 40+ test, it wasn't a test that required the liquid or CPU temp to be monitored, I didn't notice the difference in results until editing.
@@Neil3D my apologies for missing the memory timings - went through the performance graphs very quickly. It has to be EXTREMELY memory hungry app to see difference between DDR4 and DDR5 - Adobe Premiere is one such example. In specific conditions, it can gain 20%. Y-Cruncher and other scientific applications heavily depending on memory access will also benefit, but they must not be latency sensitive (bandwidth only). Some appz in SpecViewPerf benchmark suite do benefit, too. It's somehow complicated today : 99% of standard, typical applications do not benefit at all. In a year or two, when DDR5 matures and frequencies ramp up over 7000 + latencies decrease to 30-32, we will see some performance difference. Until then, just invest your money where it has real merit. You will get faster performance by using 12900k instead of 12700k or 12600k (uhh, once again not always not everywhere not all the time), but definitely not by going from DDR4 to DDR5. Expand your RAM, 64GB DDR4 3200/3600MHz (well, MT/s) costs 220-300 EUR today (2x 32GB Mushkin Essentials, Kingston Fury, Renegade, Crucial Ballistix, etc etc). That's sure a huge benefit compared to 16GB and it will be a benefit even over 32GB RAM, if you see >28GB RAM usage in Task Manager. Avoid swapping at all cost. Make sure your cooling is "more than adequate" and chassis properly ventilated (intake, outtake fans). Check your CPU/GPU temps, don't let them run over 90 degrees Celsius. Upgrade your GPUs, if working with 3D heavy stuff. Yes they are the most expensive part of workstations :( Thank you for the video and effort you put into this testing !
@@lubomirzvolensky3630 , And now, with Z790 and i7 13th generation, for example, do you think it worths it using DDR5 5200 instead of DDR4 3200? I just ordered 64 GB DDR5 for music and video production, but I am thinking about sending it back and take 64 GB DDR4 after searching more about the topic... What is your opinion?
...? I don't know what you're suggesting the difference is that I'm missing? I know the origins of Revit and was around during the Architectural Desktop days but I'm not sure what you're getting at there
By outdated you mean single core? Yeah it sure is, Useless to buy multi-cores if Revit only uses 1. 4c/8t chip is enough for it, hell I even use 4c/4t and it still does pretty well.
Greetings! I verbally provided the CL in the intro for XMP but the DDR5 is 40-40-40 1.25V and the DDR4 is 19-25-25-45 1.45V
ALSO..! If you get a blank white page on the InvMark site, there's been a BUG since day 1 with browser tracker blockers. If you have a tracker blocker i.e. Bitdefender it will cause the site to fail to load. Cadac are looking into it but its extremely difficult to diagnose. If you want to use InvMark just disable the tracker blocker for that site.
For the benefit of anyone new, the InvMark test includes a HUGE dataset of around 1000 unique files and the final score includes a grading of how the system performed whilst opening that dataset from disk on several different occasions, there's also a save to disk test, a data translation test, so InvMark isn't just a straight up plain simple "cad model a part" test by any stretch of the imagination! Also, my day 1 AutoCAD CPU index was also bang on 488 with the 12900K so I honestly don't know why the Revit results changed so dramatically, it definitely can't be thermals as that wouldn't be affected so I don't know. Still, interesting none-the-less.
Small observation made in Fusion 360 that might not have anything to do with it. NVidia studio drivers on a RTX3060 (laptop) kept it at 900MHz and even put it at 210MHz very frequently, even when trying to rotate and zoom the part. The Game Ready drivers kept it at 1425 MHz (with pushing it upwards, same as the studio drivers) and went to idle a lot less. So, maybe the Quadro driver caused the difference by keeping the card running slightly slower?
@@ExMachinaEngineering Something to investigate possibly, would be pretty bad for PR if the Quadro line was causing that much of a slow down. I'm all for longevity and reliability in the professional products but not if it means that much of a performance hit under an entry level GeForce card
Thanks for spreading useful information for professionals. There are tons of gaming tests for hardware but only few for professional application. I appreciate the inclusion of Revit even though it's not your domain.
Tom
thank you MR.tfi about all professional informations that you have given , I am really looking for your new videos every day , please keep going on ❤
Exactly what I needed to know for a necessary purchase for employment. Saves a decent amount of the painfully expensive Revit PC. It’s the equivalent of buying an HGV instead of a car. Not exotic but definitely expensive because of what it needs to be capable of
Great video, noticed very similar results with DDR4 and DDR5 testing so with current prices and shortages, I am recommending DDR4 to clients.
I always support these kind of work since everyone's testing is pointed at gaming like there is nothing else in this world except games...is it possible that nobody ask "Who's making them?"
Where I live DDR 5 Ram is currently double the price of DDR 4. Performance on both gaming and Autodesk products are the same between the 2 RAM generations. Disappointing to say the least.
I brought 2xDDR4 16g 4000 + 12700k for my revit. Works good! Run Adobe PDF very fast!!!
pdf :DDDD fuckin nonoptimised adobe! always shitty products ...
you are a menace, running such programs on a simple desktop device
Thank You...this is helpful.
I'm doing "as-built" models in Revit, which is modeling off of a pointcloud. Pointclouds can be anywhere from 6GB to 300GB. I was maxing my 32GB ram kit and getting lots of stuttering. Now on 64 and smooth as butter. If I'd build my system on a ddr5 platform (MOBO+64 GB RAM), I'd have to spend about $800 on these, while my DDR 4 (b550+4X16) was about $450. The only downside to my system is that I won't be able to have more RAM if I'll need it in the future (I have 4X16 now). So if I was building a system today, I would have gone with the DDR 5. But for most users, DDR 4 will be enough for YEARS to come.
Thanks for putting this together. It's just what I was looking for today. Have you benchmarked the integrated graphics? I don't do much 3d work in AutoCAD or Revit and wonder how much a quadro or gaming card helps 2d performance, particularly compared to Intel's latest UHD Graphics 770.
Its on the to-do list if my Z690 board is still working, there's a chance it might have been cooked by a faulty part that I was testing but if not then yep I'll run some tests on that
I don't know if this helps but i have a laptop with an AMD apu. My work is mainly 2D CAD and i found that the apu is holding good, except that it needs a larger ram than 16GB in order for a headache-free hatching on Autocad.
I’ve realized Revit doesn’t really utilize the hardware very well
I'm building a new system, 3080 ti, 12700k, dont want to spend more on ram so i will use my old 64gb of ddr4 2993mhz cl21, will it make a difference if I use that or should i get faster 3200-3600 cl16 or lower ram? I plan on running solidworks
Dude can we get a run down on 8gb vs 12gb vs 16 gb vram for revit mep?
I think ddr3 would be the same story as well. Other then PCIe 4.0 NvME support, nothing has incredibly upgraded on PC side. I mean, i9 12900k is only %50 faster then my i7 4770k (stock)? What?
DDR5 doesn't exist in the UK currently and wait time keeps adding another month with each supplier that states its on pre order
cmon! I got ddr5 16x4 = 64gb 4800Mhz for one month yet
10:16 Not going back in time to see shows a lack of commitment!
I know, I just don't have the same passion for all this as I used to! Or my time machine is out being serviced right now...
Shirt!!!!!!! Maybe save that one for the holidays 😂
It had to be said.
lets try ADOBE softwares! anyway there aren't on youtube any tests ...
Cas Latency also matters. Post those numbers.
I said what they were in the intro when explaining the test bench
@@Neil3D I must have missed it thanks
I thought the guy wasn't wearing a shirt and had a bunch of tattoos
boring huh ? Because... it was obvious :)
The thing here is, these applications simply DO NOT need and CAN NOT UTILIZE 90GB/s bandwidth available from DDR5 memory modules. Bugatti Veyron won''t be faster than Opel Astra on a road with radars/cameras where 40mph is the limit. Both will be doing 40, that's it.
Anyone running DDR4 at 3200 or 3600MHz is absolutely fine. You didn't state timings, primary and secondary, but as we see it doesn't make any difference. Uh, obviously :)
Day1 performance delta : that's not so obvious :) we guess you didn't notice any temperatures deltas on the display of liquid cooler, otherwise you would state that in video.
I wasnt monitoring the temperatures because it wasnt a day 1 vs day 40+ test, it wasn't a test that required the liquid or CPU temp to be monitored, I didn't notice the difference in results until editing.
@@Neil3D my apologies for missing the memory timings - went through the performance graphs very quickly. It has to be EXTREMELY memory hungry app to see difference between DDR4 and DDR5 - Adobe Premiere is one such example. In specific conditions, it can gain 20%. Y-Cruncher and other scientific applications heavily depending on memory access will also benefit, but they must not be latency sensitive (bandwidth only). Some appz in SpecViewPerf benchmark suite do benefit, too.
It's somehow complicated today : 99% of standard, typical applications do not benefit at all. In a year or two, when DDR5 matures and frequencies ramp up over 7000 + latencies decrease to 30-32, we will see some performance difference.
Until then, just invest your money where it has real merit. You will get faster performance by using 12900k instead of 12700k or 12600k (uhh, once again not always not everywhere not all the time), but definitely not by going from DDR4 to DDR5. Expand your RAM, 64GB DDR4 3200/3600MHz (well, MT/s) costs 220-300 EUR today (2x 32GB Mushkin Essentials, Kingston Fury, Renegade, Crucial Ballistix, etc etc). That's sure a huge benefit compared to 16GB and it will be a benefit even over 32GB RAM, if you see >28GB RAM usage in Task Manager. Avoid swapping at all cost.
Make sure your cooling is "more than adequate" and chassis properly ventilated (intake, outtake fans). Check your CPU/GPU temps, don't let them run over 90 degrees Celsius.
Upgrade your GPUs, if working with 3D heavy stuff. Yes they are the most expensive part of workstations :(
Thank you for the video and effort you put into this testing !
@@lubomirzvolensky3630 ,
And now, with Z790 and i7 13th generation, for example, do you think it worths it using DDR5 5200 instead of DDR4 3200? I just ordered 64 GB DDR5 for music and video production, but I am thinking about sending it back and take 64 GB DDR4 after searching more about the topic... What is your opinion?
@@Neil3D could you please answer my question as well? many thanks!
SolidWorks?????? 😥
ummm... isn't the result obvious ? :)
I don't even have a license of Solidworks!
@@Neil3D He probably wanted to say "Solid work you did there!" LOL
Revit itself is so old, that he does not know the difference. I wonder what would be the difference for SDRAM?
...? I don't know what you're suggesting the difference is that I'm missing? I know the origins of Revit and was around during the Architectural Desktop days but I'm not sure what you're getting at there
@@Neil3D it was a joke about how outdated revit is
By outdated you mean single core? Yeah it sure is, Useless to buy multi-cores if Revit only uses 1. 4c/8t chip is enough for it, hell I even use 4c/4t and it still does pretty well.