In my personal experience, ECC memory is crucial for working with applications that involve long and tough processes like video editing/rendering, 3d rendering, everything graphic related. I had non-ECC ram on an i5 and had crashes all the time using these applications. Now I have ECC-RAM (also on my graphics card) and my system is extremely stable and reliable. It's totally worth it.
Sounds like those crashes were due to a weak system not being able to handle what the program was asking for, leading to a long processing time which causes delays and ultimately crashes. In my personal experience, over the last year with video editing and 3D rendering, I have never had a single crash due to memory. All of the crashes were due to software and general system instabilities. I have never come across a BSOD with memory as the reason for the crash.
This is what I'm thinking. Lots of errors pop up when running CAD and other applications without closing things down. Its' time consuming to relocate and set up my working environment just to pick up where I left off
From my experience, limited though it may be, I have better luck running FEA on large models my newer machine with ECC than my older machine with non-ECC. I realize that there are other factors at play besides the RAM but with models that I tend to run overnight it's nice to come in to a completed test rather than a hung program or crashed computer.
Ya, I'm getting ecc ram on my next work pc, we are paranoid and save our work every 2 minutes, we use solidworks. I think this will help with crashes, a good graphics card helped a lot to
Just found your channel and it's refreshing to have a normal bloke who just talks like a human being and not someone amped up on energy drinks and snorting adderall. You're my type of tech channel.
you probably dont care but does any of you know of a tool to get back into an instagram account? I was dumb lost the login password. I appreciate any tips you can offer me!
6:17 From what I've read, the performance hit is due to *registered* memory (additional latency of the register), and not due to ECC itself. Unbuffered ECC memory shouldn't have any performance hit. Likewise, nothing prevents overclocking ECC RAM; it's just that the server-focused platforms that ask for ECC RAM generally don't support overclocking. On something like Ryzen, it's just as easy to overclock ECC as non-ECC RAM. Also, aftermarket DDR4-3200 ECC is readily available; it's just not offered on Dell's RAM configurator. It is a valid point that ECC RAM isn't sold at higher speeds like gaming RAM (basically guaranteed overclocking).
In short... No I use CAD (Inventor, Solidworks, Creo) and CAE (Ansys, FloEFD) software with non-ecc memory for the last decade and I never had an issue.
Unfortunately that would just be a video of me talking over product marketing pages as I'll never be provided with the hardware, routers, switches etc to demo this kind of a thing. Neat idea though, I haven't thought much about that side of things.
@@Neil3D thanks for your answer. Even If It is just you giving us some ideas would be a great vídeo! That kind of vídeo that when some one come with the Idea "lets use Wi-Fi in ower office..." We could send your vídeo explaining why not (at least thats what I think you would say kkkkk )
TLDR, If you have to ask if you need ECC ram, you don't. Not to be rude, but I feel like you omitted the largest use cases for ECC ram and I feel like I can shed a little more light on the topic. First of all, it seems like you have the misconception that processors and computer components, in general, communicate with one another without erroneous information mixed in 99% of the time. No, they do not, contrary to popular belief nearly 100% of the time communication occurs there is erroneous data mixed in, and it's not all down to background noise, is mostly due to the machine's own noise and come with the territory of the insane speeds that today's communication buses and processors operate at (to be clear you could get nearly 100% data integrity somewhere between 2 to 8 megahertz maybe, but today's processors operate in the realm of GigaHertz, and even the slowest communication buses and used today operate in the realm of hundreds of megahertz.). At this point you're probably wondering “How do computers work reliably if that's the case?”, it's simple, most software you use on your computer has some sort of mixture of parity checking and countermeasures in place for corrupt data. In addition, most of the firmware on your computer has built-in countermeasures as well. Now your question probably changed to “If that works, why do we need ECC ram again?”, all of those aforementioned countermeasures come at a cost, namely speed. Additionally, in the case of ram, to operate effectively it has to receive a message and return the identical message when requested. Some of the most common “fixes” for the issues with data reliability are added parody bits (which waste bandwidth and require extra computation.), performing the operation multiple times and comparing the results to see if they match, and the most obvious “fix” just pretend like there is no corruption and see if it works with everything else that the program has done only re-performing the calculations if it doesn't add up per se. These approaches work fine for things that are made up of many tiny elements and or that don't take too long to begin with, for example, things like web browsers, video games, and even CAD programs. In the case of web browsers if elements don't load correctly just reload the individual element, In the case of video games, things that won't make the game crash are often ignored for example, cigarettes not lining up with people’s mouths in Cyberpunk 2077, you can run the same game on the same computer and get different results each time (a lot of the times when you see stuttering, or hanging up in a game, It's generally a memory issue, especially if none of your components are overclocked.). Now let's finally talk about factors that lead to instability in ram. Hypothetically, ram gets less stable, the more you have of it, the more you use, as the total number of modules you have on each card increases, and in my experience, exponentially as you increase the number of sticks of ram per CPU. In practice, most workloads won't ever run into the instability issues that are real show stoppers that often, for example, let's say you had a system with eight gigabytes of ram, and for every 8 gigabytes you write to the ram one bit is erroneous (these aren't realistic numbers but bear with me), if you had a program that when you started it loaded 4 gigabytes into the ram and during your entire session it only needed to read about 2 gigabytes of the ram it wrote to, odds are you're never even going to get an erroneous bit performing that operation, and if you did it probably wouldn't matter. However, if your workload consisted of writing to the entire 8 gigabytes, reading the entire 8 gigabytes, performing an operation on the information, and rewriting the result to all 8 gigabytes of memory, again and again as fast as the memory will allow for multiple days and even in some cases years at a time. In this workload, you're guaranteed to see corrupted data many times per second, and what's worse is considering you're performing operations with corrupted data all future calculations using the corrupted data will be erroneous as well, so, with each clock cycle you're exponentially getting more and more corrupted useless data. This really becomes an issue when it's impractical to use some of the aforementioned “fixes” to check for erroneous data. This is exactly the kind of workload a high and virtualization server, or high-end rendering produces. In my experience with DDR4, if your rendering requires more than 32 gigabytes of ram or four individual sticks of ram whichever comes first, you're in for a bad time, if you render anything that takes more than three hours. Your renders will crash in the middle regularly unless you're using expensive high-speed ram underclocked a little bit, which at that point you could just use ECC ram. These days it's common to have 4 terabytes of ram across 16 sticks per CPU in modern servers, with 100% utilization. Believe me when I say for high-end productions it's common for workstations to have 256 gigabytes of ram for rendering workloads, and ram is often the bottleneck for many rendering operations with today's processors. a lot of the time it's unreasonable to do much error correction on large-scale renders. some movies take years to render, if you added parity bit checking for every bit it would take at least twice as long and still wouldn't catch everything, obviously if you render the entire movie three times, each copy would have different corruption, there are actually released AAA movies where CG elements just disappear and reappear randomly, and that's with all the greatest technology money can buy. Today, the best we can do is just ensure the computers don't crash while they're rendering and go through either manually or with computer vision frame by frame looking for abnormalities, re-rendering any frames containing glitches. In summary, if your workload consists of writing more than 32 gigabytes of ram as fast as the ram will allow continuously, you need ECC ram. However, if that is your workload you would probably already know that if you're not already using ECC ram. Data integrity tends to get much better with each generation of DDR ram, so I'm guessing when we get into DDR5 ram my recommendation would move up to about 90 gigabytes before you need ECC ram . If you get 64 gigabytes on eight 8 gigabyte sticks and fill it up with tabs from Google Chrome or something it will occasionally stutter and pages will crash and reload but in my testing, you won't be able to actually crash your computer. Additionally, stop pretending like AMD doesn't exist, Lenovo makes prebuilt AMD workstations, and most AMD processors, even desktop ones support ECC ram. If you actually need ECC ram support on the cheap, usually AMD is the way to go, however, if it's for an actual business unless you have on-site tech support (I don't mean you with a screwdriver even if you know everything about computers, I mean someone or preferably an entire team whose job it is to just keep the computers working) I highly recommend getting some kind of prebuilt with good customer service over building something yourself in that environment. Let me know if you have any questions, I can literally talk about this topic all day. Here's a very good video on the topic if you want to learn a little bit more about error detection. th-cam.com/video/MgkhrBSjhag/w-d-xo.html
thank you so much. you cleared up lots of stuff. i googled for long time to decide if i really need ecc ram and did not find definite answer. you provided the best answer.
I have an old Asus ATX desktop server motherboard that let you choose either kind in bios. I used it like a desktop PC with desktop memory I already had. It would be nice if all motherboards were like that.
Use autosave. Even the free open-source graphics and video editing apps have it. Then all you need to worry about is a long rendering time crashing and being wasted.
Amazing video, liked and subscribed. I was searching for a cnc kit with ECC ram ( I found a kit of acorn centroidcnc but I can't find anywhere if they include ECC RAM in that board).
1:40 That's not how it works. ECC RAM uses error-correcting codes, not parity (hence the name). Parity can detect a single-bit error but can't correct it.
Cool cat in the background, or is it? Seriously though, does this apply to what type of video ram our vid cards have onboard? In particular RTX 8000/9000 series. I’ll be having a new system built by Jet Systems in Florida, and before they do the build.... I’m wondering if I should talk to them regarding ECC RAM. Hummmm, interesting to say the least.
Quadro/Pro cards have ECC video RAM but its not something you can option for, and it's totally not something that I'd choose a GPU based on for our field. But for regular RAM, sure, have the talk with them, all the info about it is in here now and should be enough to make an informed decision based off!
@@Neil3D Thanks much for the reply, yeah I know it’s not something that can be optioned in on a video card. Was just wondering if the results would be similar, better or worse in parallel performance tests. Thanks again and stay safe. Cheers
Nothing like saving 30-50% on non-ecc ram in order to have that important drawing or rendering crash on our engineer 4 times at $60-150/hr. The Accountants and share holders don't see that and so it's all good.
I think its safe to say that if a PC is crashing that regularly then the issue isn't with bit flipping in RAM, likely internal program errors or conflicts. But... as I said in here, I always put in ECC RAM just to take that out of the conversation and eliminate it from ever being a scapegoat for someone to blame their crashes on or missing a deadline. Someone would have to be scraping the barrel to blame cosmic rays causing bit flips on their late project deliveries! But, at this point people never fail to surprise me.
It was taken from Epidemic Sounds which I'm paying for royalty free music, turns out its on TH-cam here -> th-cam.com/video/rHBLdmXtACc/w-d-xo.html Addiction - Mindme
that bothered me to. not apple to apple comparison nor price-wise (non-ecc for the same clock should always be cheaper than ecc memory), nor (clock) speed-wise.
I already answered that in the video! You can't put 4000MHz RAM into a CAD workstation, the highest is 3200MHz non-ECC RAM. I didn't explain why because it's not an "everything about RAM video" but basically when you see RAM being sold at 4000MHz speeds, that's an overclock on the RAM which is enabled in the computer BIOS. CAD workstations don't have the BIOS tools to enable that overclock. So CAD workstations can only run RAM at stock DDR4 speeds and the highest standard DDR4 speed is 3200HMz. And you (generally speaking) can't put ECC RAM into systems that would have those RAM overclocking tools in the BIOS.
@@Numian Yea but then that 4000MHz non-ECC RAM would be in a system which is completely different to a CAD workstation, it would be in a consumer grade system and then there's far too many other differences which would then outweigh the ECC vs non-ECC comparison at that point.
@@Neil3D That is true indeed. For my home PC I prefer speed/saved time over hard-to-measure reliability even I do use it for CAD/CAM and other "pro" work. Thanks.
I finished the story mode after a few days! Now there's nothing left to it, a big open world expecting me to grind and craft things with no PvE progression or target to aim for... so I sadly packed it in once the story was over.
![[MV] A leap of faith...ความเชื่อใจครั้งสุดท้าย (From "Petrichor the series")](http://i.ytimg.com/vi/U1piZH2CNXA/mqdefault.jpg)
In my personal experience, ECC memory is crucial for working with applications that involve long and tough processes like video editing/rendering, 3d rendering, everything graphic related. I had non-ECC ram on an i5 and had crashes all the time using these applications. Now I have ECC-RAM (also on my graphics card) and my system is extremely stable and reliable. It's totally worth it.
Sounds like those crashes were due to a weak system not being able to handle what the program was asking for, leading to a long processing time which causes delays and ultimately crashes.
In my personal experience, over the last year with video editing and 3D rendering, I have never had a single crash due to memory. All of the crashes were due to software and general system instabilities. I have never come across a BSOD with memory as the reason for the crash.
You had to swap out both the CPU and MOBO to get on to ECC. I reckon the issue wasn't the RAM...
This is what I'm thinking. Lots of errors pop up when running CAD and other applications without closing things down. Its' time consuming to relocate and set up my working environment just to pick up where I left off
From my experience, limited though it may be, I have better luck running FEA on large models my newer machine with ECC than my older machine with non-ECC. I realize that there are other factors at play besides the RAM but with models that I tend to run overnight it's nice to come in to a completed test rather than a hung program or crashed computer.
Ya, I'm getting ecc ram on my next work pc, we are paranoid and save our work every 2 minutes, we use solidworks. I think this will help with crashes, a good graphics card helped a lot to
Thanks for the vid, mate...working full-time in mechanical simulation / FEA, ECC gives me peace of mind when I work...my 2 cents...
Just found your channel and it's refreshing to have a normal bloke who just talks like a human being and not someone amped up on energy drinks and snorting adderall. You're my type of tech channel.
you probably dont care but does any of you know of a tool to get back into an instagram account?
I was dumb lost the login password. I appreciate any tips you can offer me!
Proud to be 3d modeler and cad operator
6:17 From what I've read, the performance hit is due to *registered* memory (additional latency of the register), and not due to ECC itself. Unbuffered ECC memory shouldn't have any performance hit. Likewise, nothing prevents overclocking ECC RAM; it's just that the server-focused platforms that ask for ECC RAM generally don't support overclocking. On something like Ryzen, it's just as easy to overclock ECC as non-ECC RAM. Also, aftermarket DDR4-3200 ECC is readily available; it's just not offered on Dell's RAM configurator. It is a valid point that ECC RAM isn't sold at higher speeds like gaming RAM (basically guaranteed overclocking).
In short... No
I use CAD (Inventor, Solidworks, Creo) and CAE (Ansys, FloEFD) software with non-ecc memory for the last decade and I never had an issue.
Sounds like a good explanation of ECC to me.
Only a detailed oriented monster can appreciate this great vidoe.
Enjoyed your post, also fitted ecc in my workstation to test out..
5:10 This is an important point that people often miss. Well said!
Video suggestion: LAN infraestructure for a CAD office.
Unfortunately that would just be a video of me talking over product marketing pages as I'll never be provided with the hardware, routers, switches etc to demo this kind of a thing. Neat idea though, I haven't thought much about that side of things.
@@Neil3D thanks for your answer. Even If It is just you giving us some ideas would be a great vídeo! That kind of vídeo that when some one come with the Idea "lets use Wi-Fi in ower office..." We could send your vídeo explaining why not (at least thats what I think you would say kkkkk )
TLDR, If you have to ask if you need ECC ram, you don't.
Not to be rude, but I feel like you omitted the largest use cases for ECC ram and I feel like I can shed a little more light on the topic.
First of all, it seems like you have the misconception that processors and computer components, in general, communicate with one another without erroneous information mixed in 99% of the time.
No, they do not, contrary to popular belief nearly 100% of the time communication occurs there is erroneous data mixed in, and it's not all down to background noise, is mostly due to the machine's own noise and come with the territory of the insane speeds that today's communication buses and processors operate at (to be clear you could get nearly 100% data integrity somewhere between 2 to 8 megahertz maybe, but today's processors operate in the realm of GigaHertz, and even the slowest communication buses and used today operate in the realm of hundreds of megahertz.).
At this point you're probably wondering “How do computers work reliably if that's the case?”, it's simple, most software you use on your computer has some sort of mixture of parity checking and countermeasures in place for corrupt data. In addition, most of the firmware on your computer has built-in countermeasures as well.
Now your question probably changed to “If that works, why do we need ECC ram again?”, all of those aforementioned countermeasures come at a cost, namely speed. Additionally, in the case of ram, to operate effectively it has to receive a message and return the identical message when requested.
Some of the most common “fixes” for the issues with data reliability are added parody bits (which waste bandwidth and require extra computation.), performing the operation multiple times and comparing the results to see if they match, and the most obvious “fix” just pretend like there is no corruption and see if it works with everything else that the program has done only re-performing the calculations if it doesn't add up per se.
These approaches work fine for things that are made up of many tiny elements and or that don't take too long to begin with, for example, things like web browsers, video games, and even CAD programs. In the case of web browsers if elements don't load correctly just reload the individual element, In the case of video games, things that won't make the game crash are often ignored for example, cigarettes not lining up with people’s mouths in Cyberpunk 2077, you can run the same game on the same computer and get different results each time (a lot of the times when you see stuttering, or hanging up in a game, It's generally a memory issue, especially if none of your components are overclocked.).
Now let's finally talk about factors that lead to instability in ram. Hypothetically, ram gets less stable, the more you have of it, the more you use, as the total number of modules you have on each card increases, and in my experience, exponentially as you increase the number of sticks of ram per CPU.
In practice, most workloads won't ever run into the instability issues that are real show stoppers that often, for example, let's say you had a system with eight gigabytes of ram, and for every 8 gigabytes you write to the ram one bit is erroneous (these aren't realistic numbers but bear with me), if you had a program that when you started it loaded 4 gigabytes into the ram and during your entire session it only needed to read about 2 gigabytes of the ram it wrote to, odds are you're never even going to get an erroneous bit performing that operation, and if you did it probably wouldn't matter.
However, if your workload consisted of writing to the entire 8 gigabytes, reading the entire 8 gigabytes, performing an operation on the information, and rewriting the result to all 8 gigabytes of memory, again and again as fast as the memory will allow for multiple days and even in some cases years at a time.
In this workload, you're guaranteed to see corrupted data many times per second, and what's worse is considering you're performing operations with corrupted data all future calculations using the corrupted data will be erroneous as well, so, with each clock cycle you're exponentially getting more and more corrupted useless data.
This really becomes an issue when it's impractical to use some of the aforementioned “fixes” to check for erroneous data. This is exactly the kind of workload a high and virtualization server, or high-end rendering produces.
In my experience with DDR4, if your rendering requires more than 32 gigabytes of ram or four individual sticks of ram whichever comes first, you're in for a bad time, if you render anything that takes more than three hours. Your renders will crash in the middle regularly unless you're using expensive high-speed ram underclocked a little bit, which at that point you could just use ECC ram.
These days it's common to have 4 terabytes of ram across 16 sticks per CPU in modern servers, with 100% utilization. Believe me when I say for high-end productions it's common for workstations to have 256 gigabytes of ram for rendering workloads, and ram is often the bottleneck for many rendering operations with today's processors.
a lot of the time it's unreasonable to do much error correction on large-scale renders. some movies take years to render, if you added parity bit checking for every bit it would take at least twice as long and still wouldn't catch everything, obviously if you render the entire movie three times, each copy would have different corruption, there are actually released AAA movies where CG elements just disappear and reappear randomly, and that's with all the greatest technology money can buy.
Today, the best we can do is just ensure the computers don't crash while they're rendering and go through either manually or with computer vision frame by frame looking for abnormalities, re-rendering any frames containing glitches.
In summary, if your workload consists of writing more than 32 gigabytes of ram as fast as the ram will allow continuously, you need ECC ram. However, if that is your workload you would probably already know that if you're not already using ECC ram.
Data integrity tends to get much better with each generation of DDR ram, so I'm guessing when we get into DDR5 ram my recommendation would move up to about 90 gigabytes before you need ECC ram
.
If you get 64 gigabytes on eight 8 gigabyte sticks and fill it up with tabs from Google Chrome or something it will occasionally stutter and pages will crash and reload but in my testing, you won't be able to actually crash your computer.
Additionally, stop pretending like AMD doesn't exist, Lenovo makes prebuilt AMD workstations, and most AMD processors, even desktop ones support ECC ram. If you actually need ECC ram support on the cheap, usually AMD is the way to go, however, if it's for an actual business unless you have on-site tech support (I don't mean you with a screwdriver even if you know everything about computers, I mean someone or preferably an entire team whose job it is to just keep the computers working) I highly recommend getting some kind of prebuilt with good customer service over building something yourself in that environment.
Let me know if you have any questions, I can literally talk about this topic all day.
Here's a very good video on the topic if you want to learn a little bit more about error detection.
th-cam.com/video/MgkhrBSjhag/w-d-xo.html
This is exactly the stuff that I wanted to find out by watching this video and what was not even mentioned there. Many thanks!
thank you so much. you cleared up lots of stuff. i googled for long time to decide if i really need ecc ram and did not find definite answer. you provided the best answer.
I have an old Asus ATX desktop server motherboard that let you choose either kind in bios. I used it like a desktop PC with desktop memory I already had. It would be nice if all motherboards were like that.
Use autosave. Even the free open-source graphics and video editing apps have it. Then all you need to worry about is a long rendering time crashing and being wasted.
Do you have a video on the whole Quadro cards for SolidWorks saga? That cost difference is bit bigger than ECC RAM.
Amazing video, liked and subscribed. I was searching for a cnc kit with ECC ram ( I found a kit of acorn centroidcnc but I can't find anywhere if they include ECC RAM in that board).
Coming to your channel to learn tricks and tips for inventor I found the actual content to be very appealing, just keep the good content, thanks
I would like to know this answer!!
1:40 That's not how it works. ECC RAM uses error-correcting codes, not parity (hence the name). Parity can detect a single-bit error but can't correct it.
.....so where was the actual ERROR correcting test? Where did you test to see if Errors actually show up from long renderings, etc?
Useful information clearing lot of things about ECC RAM. Thanks for responding to feedback. 👍
If the EU can legally force Apple to switch to USB-C. Why can’t we force manufacturers to make ecc a legal standard?
I've 2x32gb ECC ram on my dell precision 7730 with Xeon 2186, I haven't noticed any difference.
Cool cat in the background, or is it? Seriously though, does this apply to what type of video ram our vid cards have onboard? In particular RTX 8000/9000 series. I’ll be having a new system built by Jet Systems in Florida, and before they do the build.... I’m wondering if I should talk to them regarding ECC RAM. Hummmm, interesting to say the least.
Quadro/Pro cards have ECC video RAM but its not something you can option for, and it's totally not something that I'd choose a GPU based on for our field. But for regular RAM, sure, have the talk with them, all the info about it is in here now and should be enough to make an informed decision based off!
@@Neil3D Thanks much for the reply, yeah I know it’s not something that can be optioned in on a video card. Was just wondering if the results would be similar, better or worse in parallel performance tests. Thanks again and stay safe. Cheers
what you say on the AMD threadripper platform for workstations? I heard that they unofficially support ECC ram. thoughts?
Thanks mate! Another good one!!
.. does it need a specific slot-motherboard?
Omg it's the thing, thanks!!
Interesting topic! Thanks Neil.
Nothing like saving 30-50% on non-ecc ram in order to have that important drawing or rendering crash on our engineer 4 times at $60-150/hr. The Accountants and share holders don't see that and so it's all good.
I think its safe to say that if a PC is crashing that regularly then the issue isn't with bit flipping in RAM, likely internal program errors or conflicts. But... as I said in here, I always put in ECC RAM just to take that out of the conversation and eliminate it from ever being a scapegoat for someone to blame their crashes on or missing a deadline. Someone would have to be scraping the barrel to blame cosmic rays causing bit flips on their late project deliveries! But, at this point people never fail to surprise me.
if you ruining Xeon yo got to use ECC .I know thing will change but for the last 18 years I would not run anything but a Xeon in my workstation
Happy new year
HNY to you too!
Yeah but its like you said you need a Xeon chip which no regular guy would have.
Please tell me what song is playing in benchmark section :>
It was taken from Epidemic Sounds which I'm paying for royalty free music, turns out its on TH-cam here -> th-cam.com/video/rHBLdmXtACc/w-d-xo.html Addiction - Mindme
He literally compare bus 2133 and bus 2666 together? without ecc or non ecc that was too too unfair
that bothered me to. not apple to apple comparison nor price-wise (non-ecc for the same clock should always be cheaper than ecc memory), nor (clock) speed-wise.
But on desktop, you can get 4000+ MHz RAM. Wouldn't it be a different story when you would compare 2666 vs. 4400 for example? 🤔
I already answered that in the video! You can't put 4000MHz RAM into a CAD workstation, the highest is 3200MHz non-ECC RAM. I didn't explain why because it's not an "everything about RAM video" but basically when you see RAM being sold at 4000MHz speeds, that's an overclock on the RAM which is enabled in the computer BIOS. CAD workstations don't have the BIOS tools to enable that overclock. So CAD workstations can only run RAM at stock DDR4 speeds and the highest standard DDR4 speed is 3200HMz. And you (generally speaking) can't put ECC RAM into systems that would have those RAM overclocking tools in the BIOS.
@@Neil3D Yes, I understand that. I was thinking if non-ECC RAM at 4000+ MHz wouldn't outweigh potential benefits of ECC. That's all.
@@Numian Yea but then that 4000MHz non-ECC RAM would be in a system which is completely different to a CAD workstation, it would be in a consumer grade system and then there's far too many other differences which would then outweigh the ECC vs non-ECC comparison at that point.
@@Neil3D That is true indeed. For my home PC I prefer speed/saved time over hard-to-measure reliability even I do use it for CAD/CAM and other "pro" work. Thanks.
AMD supports ECC for most of the Ryzen processors.
i'am using ecc ram for my pc gaming , , , , , , , , , , ,
You enjoying cyberpunk?
I finished the story mode after a few days! Now there's nothing left to it, a big open world expecting me to grind and craft things with no PvE progression or target to aim for... so I sadly packed it in once the story was over.
You should have check that result with the same ram bus
17 minutes, some token performance benchmarks. But absolutely *nothing* real concerning error rates, error effects, or correction ability of ECC.