Nvme M5 Turbo 64gb/s

[Stoagmeyer]

So I'm buying a M.2 tomorrow (intel 600p 512g) anyway my motherboard is the z170 MSI m5. The question I've come to is that the theoretical bandwidth of the damn thing could potentially reach 64gb/s ..... well yeah i know that is in the best of what ever's and what not's . Not expecting any of that kind of numbers what ever.  The question i have is does the "twin turbo" feature what ever MSI wants to call it is that only in work in raid? Is there raid on nvme m.2?  i have a 512g coming tomorrow just wondering where this 64gb/s comes from that they are stating. Any information would greatly be appreciated

 

I apologize for my lack of umm ahh... grammar words and smart talk.. but yeah is that number in raid or can i buy 2x 512  m.2 (same brand and size) and potentially  utilize that turbo feature msi is talking about and repping up. Or would i have to raid them to use the turbo feature...tbh idk how even to turn that on.. yeah... Help Please

[Droidbot]

11 minutes ago, Stoagmeyer said:

So I'm buying a M.2 tomorrow (intel 600p 512g) anyway my motherboard is the z170 MSI m5. The question I've come to is that the theoretical bandwidth of the damn thing could potentially reach 64gb/s ..... well yeah i know that is in the best of what ever's and what not's . Not expecting any of that kind of numbers what ever.  The question i have is does the "twin turbo" feature what ever MSI wants to call it is that only in work in raid? Is there raid on nvme m.2?  i have a 512g coming tomorrow just wondering where this 64gb/s comes from that they are stating. Any information would greatly be appreciated

 

I apologize for my lack of umm ahh... grammar words and smart talk.. but yeah is that number in raid or can i buy 2x 512  m.2 (same brand and size) and potentially  utilize that turbo feature msi is talking about and repping up. Or would i have to raid them to use the turbo feature...tbh idk how even to turn that on.. yeah... Help Please

the 64Gbps (Gigabits, not bytes.) is coming from the theoretical bandwidth of the PCI-E X4 standard that NVME uses x 2 (because RAID). That's 8GBps max bandwidth. 

With MSI TT you should expect around 3000+ MBps reads and 750MBps+ writes.

It's total fucking overkill, don't really bother. You won't notice the extra speed after around 2000MBps.

[Stoagmeyer]

15 hours ago, Droidbot said:

the 64Gbps (Gigabits, not bytes.) is coming from the theoretical bandwidth of the PCI-E X4 standard that NVME uses x 2 (because RAID). That's 8GBps max bandwidth. 

With MSI TT you should expect around 3000+ MBps reads and 750MBps+ writes.

It's total fucking overkill, don't really bother. You won't notice the extra speed after around 2000MBps.

Hey thank you for the information, I installed the ssd and everything went well. I did bench mark it using userbenchmark and the R/W seem a little low, see attached for screen shot. It says its good and its average, but still quite low from what I was expecting, and alot of people have reached way better benches on rigs thought i thought would be fairly equal to mine. I fee like I'm doing something wrong. I have updated bios. 

[pyrojoe34]

34 minutes ago, Stoagmeyer said:

Hey thank you for the information, I installed the ssd and everything went well. I did bench mark it using userbenchmark and the R/W seem a little low, see attached for screen shot. It says its good and its average, but still quite low from what I was expecting, and alot of people have reached way better benches on rigs thought i thought would be fairly equal to mine. I fee like I'm doing something wrong. I have updated bios. 

I don't know that benchmark software but run CrystalDiskMark instead and show us the results (at something like 500MiB size). My guess is it's running in SATA mode instead of PCIe... although even that should give you 550MB/s read/write.

 

Edit: Nevermind, I didn't realize that was a hybrid drive... that might be normal then. It only has a 17.5GB SLC cache (the fast stuff) so if your test exceeds its capacity or has to reference a sector not in the cache, you will drop down to the slower TLC speeds. This is especially true with random reads which will suffer the most. Here's some more info for you.

[Stoagmeyer]

1 hour ago, pyrojoe34 said:

I don't know that benchmark software but run CrystalDiskMark instead and show us the results (at something like 500MiB size). My guess is it's running in SATA mode instead of PCIe... although even that should give you 550MB/s read/write.

 

Edit: Nevermind, I didn't realize that was a hybrid drive... that might be normal then. It only has a 17.5GB SLC cache (the fast stuff) so if your test exceeds its capacity or has to reference a sector not in the cache, you will drop down to the slower TLC speeds. This is especially true with random reads which will suffer the most. Here's some more info for you.

Thanks for the info, and that makes alot of sense. The website is called userbenchmark, the thing i dont understand is how some people are getting super close benches, something that looks alot closer to the advertised  1775 MB/s /  560 MB/s . Ill download crystal after windows does the cumulative update. Ill post pictures soon. 

[pyrojoe34]

12 minutes ago, Stoagmeyer said:

Thanks for the info, and that makes alot of sense. The website is called userbenchmark, the thing i dont understand is how some people are getting super close benches, something that looks alot closer to the advertised  1775 MB/s /  560 MB/s . Ill download crystal after windows does the cumulative update. Ill post pictures soon. 

It depends on where the data is pulled from and how full their cache is when they run the test. It can vary from one benchmark run to another, even restarting your PC or having open programs can effect the results with a hybrid disk.

 

With CrystalDiskMark you can choose the size of the test data to see how it affects performance. You can also use ATTO to give you a full picture at every size.

[Stoagmeyer]

1 hour ago, pyrojoe34 said:

It depends on where the data is pulled from and how full their cache is when they run the test. It can vary from one benchmark run to another, even restarting your PC or having open programs can effect the results with a hybrid disk.

 

With CrystalDiskMark you can choose the size of the test data to see how it affects performance. You can also use ATTO to give you a full picture at every size.

OK well i ran crystal, here is what it benched at also this is a fresh windows install and its just recently was update...from last post 

[pyrojoe34]

1 hour ago, Stoagmeyer said:

OK well i ran crystal, here is what it benched at also this is a fresh windows install and its just recently was update...from last post 

Now if you run it at the large size (16 or 32GiB) I'll bet it will slow down since all the cache will be used up.

 

Either way, it seems to just be an inherent property of the drive using a cache. It'll only be that fast sometimes depending on if the thing you're using is in the cache or not.