HDD vs. SSD Real World Gaming Performance @ [H]

Status
Not open for further replies.
Good info. I usually put my MP, MMO (on-the-fly texture loading is really annoying), & MOBA games only on SSD. The rest are fine on HD.

And in reality, if it fits, I put the game in RAMdisk if I know I'll be gaming for an extended session. :)
 
Bad comparison. Especially when it comes to Battlefield 4. Regardless of what bullshit you may believe BF4 can be benchmarked in multiplayer reliably. Just find some intente 64 player action on large maps and play over 10 minute stretches...from one game to the next when you look at frame times and not just silly fps you will see the SSD difference. Everytime the game has to go to the swap file or seek some file not loaded to ram that hiccup will be gone. How do I know? performance on my 6990 improved greatly once I got an SSD. I've played Battlefield since BF2 and Battlefield Franchise is a different animal in intense battles. I'm always annoyed with writers that use it to test but go about removing anything that makes the game taxing. I mean this writer went to the test range!!!

When testing with a Battlefield game you have to understand when the game is most taxing. Always all performance tests with the game say my 6990 should give me a crazy 100 fps on high settings...real work online gaming says on low-med settings i can stay about 40 fps and below 80 most of time.
 
I use a Corsair Accelerator combined with ordinary HDDs.
It's really the best of both worlds. Quick loading, quick boot up and combined with cheap storage of the conventional HDDs.:D

Too bad Samsung bought up Nvelo and brought all that to a halt.
 
5-7 years if you use a consumer SSD like a server/workstation one perhaps. I got an SSD drive recently, and by my calculation it should last me 20-30 years at least, and I'm not what you call a light user by any means.
SSD durability depends on quality related factors (TLC vs MLC chips, spare area, write amplification, OS etc) but most importantly how it's used. A typical user is not gonna stress a proper modern SSD in this area.
If, like me, you have such concerns, I would simply recommend you shop for SSDs that use MLC memory and are of decent size (256gb+) so that the spare area helps with wear leveling, anything more is overkill.
Last but not least, SSDs are good for OS/application drives, so try to use them for that and dump your media on a large slow magnetic drive.

Thanks for the usefull info!!:)
 
Testing games that don't rely on loading elements...SSD doesn't improve gameplay over an HDD, YOU DON'T SAY?!?!??!
 
Might it have been a smart idea to explain "why" the evidence adds up that way? I mean, some of us would likely have a few theories.

I was also surprised at this apparent claim that the SSD would boost framerates...why would it? It loads from the hard drive, sure, but once it's cached and/or buffered (and even a spinny hd has enough onboard memory for that at this point if that is indeed how it's handled), it's the GFX card's prerogative. I would guess the only time framerate is affected by the hard drive, is when you have major ram deficiencies?

I was nervous when I thought you guys were going to lump loading in there with that too. After all, I figured based on how SSD's work that they would simply have to be a ton faster with anything that actually has to do with immediate hard drive usage.

Anyway yeah, an explanation as to why we do and do not get the benefits and differences seen may have been nice? I feel understanding a thing is better than any one writeup or benchmark.
 
As we all know, moving to SSD improved your productivity in installation and load time. That would make it worth while to upgrade IMO.
 
One of the measurable performance problems with an HDD is repeated frames. When a game engine waits for data from an HDD it will display the same frame repeatedly, creating that stuttering feeling. Intel did measurements of this, and found huge differences.
They never claimed it to increase framerates, SSD's merely reduce or eliminate the number of duplicate frames. In order to measure this, each frame needs to be recorded and compared to each other.
If you have 60 FPS, but chunks of those are similar frames, you may actually only see 40 NEW frames, for instance. You are still at 60, but groups of frames are repeated.
 
Also, game engines load 'filler' textures and shaders into the VRAM of the GPU for those times when they cannot receive information from teh storage subsystem fast enough.
In a racing game, for instance, if there is a lag in receiving textures as you race around the track, substitute fillers will be used instead. This keeps the game from stuttering, but definitely impacts the quality of the game.
If you fall off a cliff in the game you might only see a blur of the cliff face with an HDD. With an SSD you might see actual pebbles, etc, as they shoot past.
The storage can deliver the information faster to the game engine, thus there is no use of substitute data that is of low quality.
It is important to remember that game developers code everything in the game to give the HDD a 'crutch' so that its slow performance doesn't impact the overall gameplay. However, there is a trade-off to keep the game from playing as stutter-free as possible.
 
I never saw any benefit while running an SSD, so I switched back to big ass spinners. SSD is great in a laptop, but feels silly in a desktop for gaming.
 
I have a bit of advice for someone that is new to SSDs and just recently got a cheap drive: Chances are, you'll run into a bit of a "stutter" issue from time to time, especially if you are using the SSD as an OS drive. When the issue occurs, the computer still responds, but you can't get any response from anything that requires data from the SSD. This will occur a lot on cheaper SSD drives when you try and write a certain amount of data to it. To work around it, change all of your browser and Windows temp folder to a secondary HDD. That saved me a lot of aggravation.
 
I've been arguing with people for years that SSD's don't improve gaming hardly if at all. Thanks for taking the time to confirm :)
 
Also, game engines load 'filler' textures and shaders into the VRAM of the GPU for those times when they cannot receive information from teh storage subsystem fast enough.
In a racing game, for instance, if there is a lag in receiving textures as you race around the track, substitute fillers will be used instead. This keeps the game from stuttering, but definitely impacts the quality of the game.
If you fall off a cliff in the game you might only see a blur of the cliff face with an HDD. With an SSD you might see actual pebbles, etc, as they shoot past.
The storage can deliver the information faster to the game engine, thus there is no use of substitute data that is of low quality.
It is important to remember that game developers code everything in the game to give the HDD a 'crutch' so that its slow performance doesn't impact the overall gameplay. However, there is a trade-off to keep the game from playing as stutter-free as possible.

^
|
This! Not to mention this article seemed rushed. I read a lot of copy/paste in this article, from video card reviews. While I am all for shortcuts, it really should be altered to reflect the subject in discussion.

Personally, I consider the ability to save time zoning/loading/entering games faster a "gameplay performance improvement." When did FPS == THE be all end all of gamplay performance. Especially, in competitive environments that need a faster responses to events. I cannot begin to say how many times I have had a bad night raiding/gaming/pvp when my team mates had crappy HDD (4900/5400rpm) and could not load textures causing death or wipes for the team.
 
For those looking to see how using a SSD affects playing WoW. Here is loading times video i made some time ago. The loading screen is a little quicker but not that big a deal. But notice after the loading screen all the players just seem to 'appear' instantly instead of slowly poping in. After using a SSD for wow I always have it installed on a SSD. Other games go on my normal HDD.

http://youtu.be/qa6FpWkk9qo
 
*sigh*

Everything done here was WRONG.

Remember first, that most games aggressively pre-load most of the data they need at any point in time, so for most situations, you won't have a major impact once a level is loaded. At that point, HDD impact is already minimal. However, the way this was tested is WRONG.

*) How do HDD's affect gameplay performance? When data needs to be read off them.
*) When does data get read of HDD? When Paging occurs.
*) When does paging occur? When you run out of free RAM.
*) How much RAM was used in the test system? 16GB.

Reduce to, say, 4GB of RAM, and bench a game that actually uses more then 2GB of RAM [This requires it be compiled as Large Address Aware, otherwise the Win32 2GB restriction on Address Space usage remains in effect], and you'll get a different result set then what you see here, as reading from the HDD will occur more often.
 
Load heavy open world games like Skyrim, Fallout 3, and the Mass Effect series are a completely different experience on an SSD. They're far more immersive without the endless loading screens. If you play those kinds of games, an SSD is a must have. This article completely misses the point of why someone would want an SSD for gaming. It doesn't even test the right kind of games. Put the multiplayer shooters aside for a moment and go try some open world games and tell me an SSD doesn't make an enormous difference in the experience.
 
Last edited:
good write-up, though I think most of us ([H]-informed) readers upgrade to SSD for read/write speed, not potential FPS gains.

I think the next area to tackle is TRIM Vs RAID. If TRIM assists a SSD with deleting blocks, preventing the need for defraging, but RAID 0 provides the best speed... what is the trade off?
 
dunno if its SSD but I usually am one of the first to load into a BF4 map... Get my pick of assets and I am usually out of the main base before other people load.

So single player games loading doesn't mean much but multiplayer loading can be huge.
 
Load heavy open world games like Skyrim, Fallout 3, and the Mass Effect series are a completely different experience on an SSD. They're far more immersive without the endless loading screens. If you play those kinds of games, an SSD is a must have. This article completely misses the point of why someone would want an SSD for gaming. It doesn't even test the right kind of games. Put the multiplayer shooters aside for a moment and go try some open world games and tell me an SSD doesn't make an enormous difference in the experience.

Skyrim cranked up to max with some hd textures runs fine on my spinner.
 
Skyrim cranked up to max with some hd textures runs fine on my spinner.

Sure, but how are the load times when moving between areas?

I remember Oblivion being practically unplayable due to the load times on some slower HDDs. Can't really speak for Skyrim because I had an SSD by the time it came out.
 
Good information for sure. But I'm not sure if I ever was under the impression that SSDs = better gameplay performance. It's always strictly been about loading times/transitions, from what I recall.
 
*sigh*

Everything done here was WRONG.

Remember first, that most games aggressively pre-load most of the data they need at any point in time, so for most situations, you won't have a major impact once a level is loaded. At that point, HDD impact is already minimal. However, the way this was tested is WRONG.

*) How do HDD's affect gameplay performance? When data needs to be read off them.
*) When does data get read of HDD? When Paging occurs.
*) When does paging occur? When you run out of free RAM.
*) How much RAM was used in the test system? 16GB.

Reduce to, say, 4GB of RAM, and bench a game that actually uses more then 2GB of RAM [This requires it be compiled as Large Address Aware, otherwise the Win32 2GB restriction on Address Space usage remains in effect], and you'll get a different result set then what you see here, as reading from the HDD will occur more often.

The proper response to a low RAM scenario is not reduce paging delays with buying faster storage, it's buy more RAM.

Lots of games stream data off the drive while you move around instead of the 'loading screen loads everything it needs for this level' model. Which is why the games that should have been tested here should be more MMO style or open world/procedural games rather than the games specifically designed not to have load delays while playing it in little consumable sections.
 
Back when I was in a Red Orchestra 2 clan, and played the game a ton, I found that when I played on pubs, the only benefit a fast drive (like an SSD) had was getting into the game fast after a map change, and having the choice of classes/weapons before the most popular ones were taken.

I even threw 32Gigs of ram in my desktop so I could create a 16gig RAM Disk in order to speed this up as much as possible.
 
Like many have already said this would have been better tested using games which actually DO access the drive during game play.....also frame time variance would be a far better indicator of improvement than simple average FPS.

I wish the industry would get to grips with the ssd/hdd issue.....caching technologies are part of the way there for increasing boot speeds but of course don't help for those one off data loads.

I think a company called Superspeed (?) had a good tech, with their Super Volume which enabled you to mirror a drive or partition in ram.

Now if those two techs could be merged to enable caching on an ssd and for user selected program files, such as games, to also be mirrored on the ssd then that would be just about a perfect solution.

I do something similar to this with my steam games having them installed on a large hdd but using a "steam mover" application to move them onto a small ssd for game play.....although don't currently have a hybrid caching hdd so my OS resides uncached on the hdd :-/
 
Last edited:
Like many have already said this would have been better tested using games which actually DO access the drive during game play.....also frame time variance would be a far better indicator of improvement than simple average FPS.

I wish the industry would get to grips with the ssd/hdd issue.....caching technologies are part of the way there for increasing boot speeds but of course don't help for those one off data loads.

I think a company called Superspeed (?) had a good tech, with their Super Volume which enabled you to mirror a drive or partition in ram.

Now if those two techs could be merged to enable caching on an ssd and for user selected program files, such as games, to also be mirrored on the ssd then that would be just about a perfect solution.

Doesn't expresscache already do something similar to this? I can preload directories or files from the command line and when I do my cache stats I get a chart like this:

Code:
EC Cache Info
==================================================

Mounted                   : Yes
Partition Size            : 16.00 GB
Reserved Size             : 3.00 MB
Volume Size               : 16.00 GB
Total Used Size           : 14.19 GB
Total Free Space          : 1.81 GB
Used Data Size            : 14.12 GB
Used Data Size on Disk    : 14.18 GB

Tiered Cache Stats
==================================================
Memory in use             : 960.00 MB
Blocks in use             : 7535
Read Percent              : 9.27%


Cache Stats
==================================================
Cache Volume Drive Number : 1
Total Read Count          : 97131
Total Read Size           : 2.34 GB
Total Cache Read Count    : 35849
Total Cache Read Size     : 1.09 GB
Total Write Count         : 79199
Total Write Size          : 1.28 GB
Total Cache Write Count   : 23258
Total Cache Write Size    : 342.50 MB

Cache Read Percent        : 46.29%
Cache Write Percent       : 26.22%
 
That's interesting Traciatim, I've no actual hands on experience of caching tech.....so that allows you to preselect a given drive partition to be cached on the ssd as well as caching frequently used files?....would be better though if you could select individual files or folders to be cached so you could select individual game program folders......that way you wouldn't have to "move" the game folder to a smaller partition for caching otherwise you would have to cache the all your games installed in the main drive or partition and the ssd would not be big enough.
 
That's interesting Traciatim, I've no actual hands on experience of caching tech.....so that allows you to preselect a given drive partition to be cached on the ssd as well as caching frequently used files?....would be better though if you could select individual files or folders to be cached so you could select individual game program folders......that way you wouldn't have to "move" the game folder to a smaller partition for caching otherwise you would have to cache the all your games installed in the main drive or partition and the ssd would not be big enough.

It caches a whole drive, it tracks usage and your common stuff gets automagically moved on to the SSD. As you can see, since this is my work laptop and today was mostly working away from my desk or in RDP on another machine, near 50% of my reads are off the SSD for the day and it's only 16GB.

It has a command line interface where I can force it to cache specific folders or files on the drive, but I generally don't use it since I figure the stuff I use most will flow in and out of the cache as needed.

I also use Intel SRT on my home machine with a 60GB SSD. When I first installed it and ran some tests I couldn't tell the difference between the OS installed directly on the SSD or just using it as a cache and having everything on my spinning disk. I'm sure now that I have a few hundred GiB of junk on my machine it might be different now, but I've been really impressed by the SSD cache setups in both cases.

On a great feature to caching note, I don't have to manage which apps and data go on which drive, it just ends up that the stuff I use gets faster the more I use it. I never really understood why people want their whole OS on there, it just seems to me that's a waste of space since I would think well over half of the windows directory doesn't get accessed except in very rare cases.
 
It caches a whole drive, it tracks usage and your common stuff gets automagically moved on to the SSD. As you can see, since this is my work laptop and today was mostly working away from my desk or in RDP on another machine, near 50% of my reads are off the SSD for the day and it's only 16GB.

It has a command line interface where I can force it to cache specific folders or files on the drive, but I generally don't use it since I figure the stuff I use most will flow in and out of the cache as needed.

I also use Intel SRT on my home machine with a 60GB SSD. When I first installed it and ran some tests I couldn't tell the difference between the OS installed directly on the SSD or just using it as a cache and having everything on my spinning disk. I'm sure now that I have a few hundred GiB of junk on my machine it might be different now, but I've been really impressed by the SSD cache setups in both cases.

On a great feature to caching note, I don't have to manage which apps and data go on which drive, it just ends up that the stuff I use gets faster the more I use it. I never really understood why people want their whole OS on there, it just seems to me that's a waste of space since I would think well over half of the windows directory doesn't get accessed except in very rare cases.

So the expresscache has the ability to specify files and folders via command line as well as general caching of frequently used stuff......that sound just like what I want.

Does Intel SRT also have the ability to specify individual files/folders for preload?

Is expresscache purchaseable stand alone software or bundles with specific ssd's?
 
I have to wait to read this until I'm home, but I originally had BF4 on my spinner and loading maps took in excess of 45-60 seconds. I stuck it on my Samsung 830 after I got an 840 EVO for a boot drive and now I can load in around a third of the time. For game modes where the first people to load in will get the first objectives this is pretty important to me, but I've never seen SSD do anything for frame rates except in a rare few games that have extremely poor texture streaming and thrash the hard drive badly.
 
LOL, the guys freaking out about systems not tested with 2 and 4GB of RAM. Come on. DDR3 is pretty cheap, even for 8GB. Forcing the system to page anything accomplishes nothing other than showing you could use more RAM. For any game that relies on loading the bulk of it's data into ram at once, an SSD will make that part faster, it has nothing to do with framerates. For a game that relies on streaming everything off of the drive constantly, it could possibly help, but even that should only resolve some performance dips or things like texture pop in, not a boost in overall framerate.

Even if using an SSD in a system with 2GB of ram did make a difference, you'd still want more RAM to improve things anyway, so at that point your priorities are simply screwed up.
 
I have a bit of advice for someone that is new to SSDs and just recently got a cheap drive: Chances are, you'll run into a bit of a "stutter" issue from time to time, especially if you are using the SSD as an OS drive. When the issue occurs, the computer still responds, but you can't get any response from anything that requires data from the SSD. This will occur a lot on cheaper SSD drives when you try and write a certain amount of data to it. To work around it, change all of your browser and Windows temp folder to a secondary HDD. That saved me a lot of aggravation.

This is one of those things that I did not know, but am always thinking about when I ponder if I should make the switch (by which I mean use an SSD for my OS drive + other things...they're still a bit pricey in my opinion to have two...). Of course many users on this very forum over the last year or so have at least convinced me that the slowdown/total life of the thing is not something to realistically be concerned about at least..

*sigh*

Everything done here was WRONG.

Remember first, that most games aggressively pre-load most of the data they need at any point in time, so for most situations, you won't have a major impact once a level is loaded. At that point, HDD impact is already minimal. However, the way this was tested is WRONG.

*) How do HDD's affect gameplay performance? When data needs to be read off them.
*) When does data get read of HDD? When Paging occurs.
*) When does paging occur? When you run out of free RAM.
*) How much RAM was used in the test system? 16GB.

Reduce to, say, 4GB of RAM, and bench a game that actually uses more then 2GB of RAM [This requires it be compiled as Large Address Aware, otherwise the Win32 2GB restriction on Address Space usage remains in effect], and you'll get a different result set then what you see here, as reading from the HDD will occur more often.

Yeah, I think this sums up what the review conclusion explanation was at least lacking. At least better than my past post.
 
*sigh*

Everything done here was WRONG.

Remember first, that most games aggressively pre-load most of the data they need at any point in time, so for most situations, you won't have a major impact once a level is loaded. At that point, HDD impact is already minimal. However, the way this was tested is WRONG.

*) How do HDD's affect gameplay performance? When data needs to be read off them.
*) When does data get read of HDD? When Paging occurs.
*) When does paging occur? When you run out of free RAM.
*) How much RAM was used in the test system? 16GB.

Reduce to, say, 4GB of RAM, and bench a game that actually uses more then 2GB of RAM [This requires it be compiled as Large Address Aware, otherwise the Win32 2GB restriction on Address Space usage remains in effect], and you'll get a different result set then what you see here, as reading from the HDD will occur more often.
Even though I also commented on 4GB, I think most gaming systems have more than that. Also, artificially creating a scenario where disk performance is the bottleneck (reducing ram) is unfair like cranking up resolution to show that a specific GPU isn't very good.

Disk performance really is a game by game basis. I don't think this [H] writeup tells us SSD's do not help games, I just think it tells us that SSD's do not help the games they tested.
 
I would have been surprised if this would have impacted FPS. Having had SSDs for a few years, I do appreciate the faster load times. In some of the older multi-player games you really had a big advantage by getting in faster. It seems to me tho that for most of the newer games this has been negated by a delayed start for each new map.
 
Status
Not open for further replies.
Back
Top