I know, it was merely a test. Some games will always crash without one, some do not, it depends how they are programmed.
Last edited:
I found out what was freezing Watch Dogs: Legion (can see it in task manager)
- Thread starter g67575
- Start date
It's possible to just install more RAM tbh (but expensive), I don't think the game accessing the page file itself (regardless of which drive it's on) actually matters that much, the game files are what caused the gameplay freezing on my SSD.
I need to test the game a bit more to make sure the stuttering isn't happening after defragging the game files, but I played for a couple of hours last night and it was fine. One possible caveat, is that if the game files are changed with a game update, some of the files may need to be defragmented again.
Last edited:
Yea but once a game is installed run defrag to trim the drive and things should be fine.
Games not gonna do constant writing its usually just read only after installing.
No software knows about page file.
It's Windows which handles allocating memory to programs also handling (or mishandling) page file use.
And speaking of that, on which drive you have page file?
You mentioned OS drive being small.
So if also page file is on that BX500 maybe that mixed simultaneous load could cause some performance issue on what stinks like QLC drive, or at least gargabe bin TLC.
Soldato
This is why I suggest he check to see what else the OS is doing when this happens because this, if it's accurate, is actually garbage:
BX500 vs my own 850evo.
Oh and the reason you thought the drive had a ~200ms read access time is because you read the benchmark results wrong, g67575. It's 0.254ms, not 254ms
No they're not, even the fastest modern day HDD can only hit 87 MB/s, that's at least 10 MB/s slower than a QLC drive. Yes it slows down once the SLC cache is exhausted but it's still faster than a HDD.
The freezing still seems to be happening occasionally unfortunately...
I've noticed that after the freezing occurs, the game loads more of the game files into the physical memory, reported as 'in use' in the Windows Resource Monitor. Initially, the game loads around 12GB, but it can rise to 13GB in some cases. I let the developers know about this SSD related freezing issue a few days ago, I guess there's a chance they might patch it still.
Also, the game gradually allocates a lot of RAM (5GB in my case) to 'standby' memory, where it tends to remain.
I've noticed that after the freezing occurs, the game loads more of the game files into the physical memory, reported as 'in use' in the Windows Resource Monitor. Initially, the game loads around 12GB, but it can rise to 13GB in some cases. I let the developers know about this SSD related freezing issue a few days ago, I guess there's a chance they might patch it still.
Also, the game gradually allocates a lot of RAM (5GB in my case) to 'standby' memory, where it tends to remain.
Last edited:
Cannot help feel ssd access times are the wrong direction, update bios, reinstall windows update all drivers the issue will likley go away.. (about an hours work)
Slowest to fastest ssd data transfer rates, even fastest nvme to slowest (modernish) sata is barely 50% faster in real world home based applications.
Even moving from hdd to ssd barely gives a 2 or 3x increase in game load speeds (best case)
Slowest to fastest ssd data transfer rates, even fastest nvme to slowest (modernish) sata is barely 50% faster in real world home based applications.
Even moving from hdd to ssd barely gives a 2 or 3x increase in game load speeds (best case)
That link shows a lot more than 87mb/s sequential write for the mechanical drive. Infact more than double
Not sure what graph you're looking at but it should be the last one as all the others don't show sequential writing, unless you consider read/writing a 2Mb file to be actually reading or writing to the disk when i don't think there's a single HDD currently being sold that doesn't come with more than a 2MB DRAM cache.
Soldato
The disk capture benchmark is multiple reads and writes over a set period of times. It's not a sequential benchmark.
Doesn't matter, a single large file still wouldn't be faster than a QLC drives after running out of SLC cache.
In other words this comment...
In other words this comment...
Is simply wrong.
Soldato
Well, debatable, but the point was you're trying to compare a mixed multiple read/write benchmark to a concurrent write only benchmark to show HDD's are still slower when they are two very different workloads (and the mixed benchamrk is MUCH harder on spinners of course..). It makes zero sense to compare them. Is that drive even up there with the fastest mechanicals any more? It's 6 years old now...
BS.
That's Donald Trump and Vladimir Putin load of BS!
Best HDDs reached 150 MB/s STR decade ago in early part of the drive on outer cylinders.
Now even 5400 RPM drives exceed ~200 MB/s in outer cylinders averaging around 150 MB/s for over the drive.
And 7200 RPM drives do up to 250 MB/s averaging near 200 MB/s.
https://www.servethehome.com/wp-con...Ultrastar-DC-HC510-10TB-HD-Tune-Pro-Write.jpg
https://www.servethehome.com/hgst-wd-ultrastar-dc-hc510-10tb-sata-hdd-review/3/
Doubt that BX500 of thread's starter averages that much over 100 MB/s for whole drive write.
And Samsung 870 QVO averages below 100 MB/s just in 15 minutes long ~87GB write.
https://www.techpowerup.com/review/samsung-870-qvo-1-tb/6.html
Any even remotely modern HDD would finish writing 100 GB far before that.
Associate
You need to get your head around what the pagefile is for and how it is used by Windows.
If the system is running low on physical memory, Windows will start to move pages out of physical memory into the page file. Which is why your bizzarre experiment with putting the pagefile on a RAM disk was doomed to failure - Windows started to move pages to the pagefile because it was low on physical memory, but because the pagefile was also in memory that just meant that it actually used more physical memory, so it had to free more physical memory and so on. If the system runs out of virtual memory (physical+pagefile) then Windows will start terminating processes which is exactly what then happened.
No application program needs a pagefile to be present. If there is no page file then all that happens is that an attempt to allocate more memory than there is physical memory will fail and the program will probably crash. But that is not because of the lack of pagefile, as the same can happen if the program attempts to allocate more than the sum of the physical memory and the space allocated to the pagefile.
In general, Windows memory management does work better if there is a pagefile, though; and for most desktop usage, just leaving the settings alone and letting Windows manage it is absolutely the best approach.
So, what happens here is that when the game loads and requests a big chunk of memory Windows will start to page out other processes. Page out means it will write the contents of the physical memory pages from those applications into the page file so that the physical page can be used by the game instead. That's the activity you are seeing against the page file not anything to do with the game itself. The other thing which will happen is that Windows will shrink the amount of memory it's using for filesystem cache which can also increase the amount of I/O going on.
If those applications are themselves active, then Windows will at some point have to page them back in - i.e. read the pages back from the pagefile into physical memory. To do that it may have to page out part of the game's memory.
You need to look at what else you are running in the background to ensure you are not over-committing your physical memory.
No, there is no debate. A QLC drive is simply not slower than even a modern HDD, let alone one from 10 years ago.
Nice manners you have there.
Does insulting people often convince people you're right?
I call things what they are.No, there is no debate. A QLC drive is simply not slower than even a modern HDD, let alone one from 10 years ago.
Nice manners you have there.
Does insulting people often convince people you're right?
And you sir are a liar and yet again spreading BS!
83MB/s vs 143MB/s average STR:
The 1TB 860 QVO falls below the sequential write speed of a 1TB hard drive once the SLC cache runs out
https://www.anandtech.com/show/13633/the-samsung-860-qvo-ssd-review/2
And that was against propably nearing 10 years old HDD...https://www.anandtech.com/show/13633/the-samsung-860-qvo-ssd-review/2
Crappy QLC Flash simply needs massive amount of parallelism to match "old spinning rust".
So that's a no then!
I'll tell you what, you can carry on believing HDD are better than QLC SSD and I'll carry on thinking the opposite, because there's little point in discussing this with someone who can't even keep a civil tongue in their head.
Soldato
Try it on another drive.