SSD Performance In Crysis 2, World Of Warcraft, And Civilization V

[torque79]

Good article. It does show that the benefits are largely in reducing our need for patience waiting for game loads. Though I'd love to get an SSD, I'd rather buy a better video card for now and buy an SSD when the price goes down further, hopefully when the next capacity tier comes out.

 

[cadder]

I patiently read through all of the "dry benchmark stuff" waiting for some real results, and there were none! Send the writer back to the closet and don't let them out until this article is finished.

First off you say "But what do input/output operations per second and megabytes per second really mean to the enthusiast".

Then you give a big analysis of mainly "que depth" for each title.

Then you end with a big chart that mostly says "sequential write performance" generated by a benchmark program.

I didn't find anything that would help a person determine how much they could speed up any of the games with an SSD. And you never answered the original questoins. Why don't you guys at least buy a cheap stopwatch and at least do some crude timings with it?

 

[heroictofu]

I agree with a lot of the comments here. It wasn't very clear how an SSD actually compares to a HDD. There were no actual comparisons in the graphs during each test phase which doesn't really help establish what the SSD is doing different/better/worse.

 

[cpy]

I bet my wow with tons of addons, custom sounds and icons, with lot of saved auction data, makes hell on plate for my hdd
I bet it would tear down ssd!

 

[Guest]

What you rEALLY need to test is how RAMDISK loading with 12-24 GB compares to SSD performance on a price/performance ratio.

RAM may be even cheaper than SSDs for better performance! Just load all your game files on RAM DISK with a good prog and see what happens?!

 

[-Fran-]

Did you do the "Orgrimar test", Andrew? I just built a RAID 0 and there is little to no frame drop while going around the city at all when loading the textures. It's really nice.

Cheers!

PS: Good to see you're Horde!

 

[wooodoggies]

so if i have a barracuda, it doesnt seem like i need an ssd according to the last chart???

 

[jsowoc]

The article is a good introduction, but I think what many readers were hoping for is:

A) I have a HDD and it takes the game/level 56.7s to load; how much faster would this be on a SSD?

B) I have a HDD and I sometimes get stutter as the game writes its updated state; how much would the minimum FPS change (if at all) if I upgraded to a SSD?

Could you write a follow-up article addressing the above two points?

 

[acku]

So it looks to me like game loading and level loading is not significantly hard-disk bound, if the disk is busy for such a short period of time. For example, loading a Crysis 2 level taking 58s, of which the disk is busy for 2.

Does that mean if you had an infinitely fast disk, the level loading would take 56s? In which case, where is the bottleneck for level loading? Is it CPU bound? (if so, why isn't CPU usage at 100% when loading a level?) Memory? Graphics card?

Thanks for the feedback everyone. SSD reviews typically provide information on the "four corners test." (4K random and 128K sequential read/write). The point of this article to give you an idea of how games relate to testing.

I, Anand, and others have commented on the real-world aspect of SSDs. There's a limit to SSD performance, where adding a faster and faster SSD won't cut down your bootup, game loading, or level loading time. That's what PCMark 7 reflects in the Game Loading test (which is technically identical to our Wow game loading trace here). And if you're looking for a HDD to SSD comparison, I would point you in the direction of one of our recent SSD roundups, because that benchmark has already been done.

But to answer your question, an infinitely faster disk won't give you unbelievable performance. There is a limit to real-world benefits, which is what PCMark 7 is designed to show.

That's what you're seeing here. Even on a very fast 240 GB Vertex 3, the disk busy time is like 2 secs when you load Crysis 2, but total game loading time is ~30 secs. Why? Because the system is busy doing things beyond querying the disk for data. There is CPU processing, loading data into memory, loading into CPU cache, loading GPU textures, etc... etc...

That won't change much when you downgrade you go to a 64 GB m4. Busy time may be +3 secs but the overall effect on game loading isn't going to change very much beyond say 33 seconds. Compared to a HDD, there's a world of difference, but very little between SSDs when you look at one specific case.

However, that doesn't tell the full story, because moving up to a faster SSD does help system responsiveness when you look at the BIG picture. If you were to measure disk responsiveness over the course of a week doing different tasks, you will feel the total effect of having a faster SSD. That's the point of using a long trace, specifically our Storage Bench v1.0. It provides an actual indicator of storage responsiveness from disk to disk over the long-term.

Cheers,
Andrew Ku
TomsHardware.com

 

[KelvinTy]

What about using SSD as a cache for the large HDDs?? and what would it look like if you compare it to raid 0 stripping drives?
Thanks.

 

[acku]

i like how tom's deletes any posts that prove test procedure failures and clear "advertorial" reviews

As far as I can see, all of the critical posts have been preserved. Everyone is here to learn more, right? That's the idea of any review site. I understand you might have frustrations about a particular test methodology, but I would suggest that a dialog might be more beneficial. We're pretty even handed, especially considering some of the nastier emails that I randomly get.

What about using SSD as a cache for the large HDDs?? and what would it look like if you compare it to raid 0 stripping drives?
Thanks.

I suppose that depends on the caching. Write back or write through (enhanced mode for Z68). Write through will give you the boost for game and level loading but gameplay will be about the same.

 

[larkspur]

[citation][nom]icepick314[/nom]I imagine in real world situation, 2 of 10000RPM HDDs in RAID 0 are just as fast as 2 of the most SSDs in RAID 0....[/citation]

No, they aren't. Go back in the article and read the theory sections. It is spelled-out why mechanical disks (even the faster spinning ones) lag behind in several real-world I/O categories. SSDs scale better in RAID-0 as well.

 

[S2Hizzle]

Andrew, you said "Compared to a HDD, there's a world of difference, but very little between SSDs when you look at one specific case."

It would be great to have another chart added to the article with comparison load times between a HDD and a SSD. I don't need a comparison of a hundred different SSDs, just a baseline HDD benchmark and the improved performance with an example SSD. While I appreciate the analysis of what the drives are being tasked with, it doesn't necessarily give the reader the data they're looking for. Does it cut the load time from 2 minutes to 30 seconds, or merely 35 seconds down to 30 seconds? In gameplay, does a faster drive have any benefit to framerates? That's the data that I and I think a bunch of the other readers that left comments are looking for.

 

[cangelini]

i like how tom's deletes any posts that prove test procedure failures and clear "advertorial" reviews

I'm able to see all moderation actions taken in each forum thread. Not a single post has been deleted in this story. Moreover, this story was motivated by a wish to see how SSDs affect various aspects of gaming, period. Please check your facts before making accusations. Consider this your warning.

Thanks,
Chris

 

[acku]

Andrew, you said "Compared to a HDD, there's a world of difference, but very little between SSDs when you look at one specific case."

It would be great to have another chart added to the article with comparison load times between a HDD and a SSD. I don't need a comparison of a hundred different SSDs, just a baseline HDD benchmark and the improved performance with an example SSD. While I appreciate the analysis of what the drives are being tasked with, it doesn't necessarily give the reader the data they're looking for. Does it cut the load time from 2 minutes to 30 seconds, or merely 35 seconds down to 30 seconds? In gameplay, does a faster drive have any benefit to framerates? That's the data that I and I think a bunch of the other readers that left comments are looking for.

I think that's a reasonable request. Maybe like a round 2.

Though, I can tell you right off the bat that it's not going to improve frame rates. We would have to artificially create a situation where the hard drive was heavy fragmented in order for the disk to load textures slower. That would affect frame rates. Think of the stuttering that occasionally happens when you're gaming. Translating that part of gameplay into something reproducible in the lab is a bit of a head scratcher.

Cheers,
Andrew Ku
TomsHardware.com

 

[buzznut]

[citation][nom]jacobdrj[/nom]Meh, the faster I can start Fragging, the better...But while I can NOT live without an SSD as a boot drive, I CAN live with my games being on a 3-way RAID-0 of 10,000 RPM Raptors...[/citation]

If you're willing to spend the extra for performance than why not an SSD? People without SSD's are skeptics, people with SSDs are BELIEVERS.

 

[cypeq]

You need some balls to call 7.1 on 4.1 (per ear) mini speaker setup

 

[cypeq]

[citation][nom]cypeq[/nom]You need some balls to call 7.1 on 4.1 (per ear) mini speaker setup[/citation]
shit not here ;D

 

[11796pcs]

They should have used a 470/570/6950/6970 or something more powerful. The whole point of the article was to benchmark performance of the SSD but you probably bottlenecked your system through the subpar graphics card. Get a 6990 or something ridiculous do that your results are proven.

 

[acku]

They should have used a 470/570/6950/6970 or something more powerful. The whole point of the article was to benchmark performance of the SSD but you probably bottlenecked your system through the subpar graphics card. Get a 6990 or something ridiculous do that your results are proven.

Look.... That's really not what's going on here. When you execute a game, you're not rendering anything. The CPU still has to load the data into RAM and run the code. There is lot of information available if you want to know why this is. I suggest reading up on tiered storage if you want more detail. In a nutshell, there is latency on the bus for that. I could have gone quad SLI and that still wouldn't change anything. We could easy have built a $4500 system completely HPC with some crazy fast parts, but that still wouldn't make game loading or level loading instantaneous. Nor would it improve frame rates, because everything is in RAM at that point.

 

[bizzzorro]

[citation][nom]AlexIsAlex[/nom]So it looks to me like game loading and level loading is not significantly hard-disk bound, if the disk is busy for such a short period of time. For example, loading a Crysis 2 level taking 58s, of which the disk is busy for 2.Does that mean if you had an infinitely fast disk, the level loading would take 56s? In which case, where is the bottleneck for level loading? Is it CPU bound? (if so, why isn't CPU usage at 100% when loading a level?) Memory? Graphics card?[/citation]

Agree this is something you suppose to see in this article. If I want to speed up the launching and gameplay, SSD´s are not the right choice, because the bottleneck is .... PLEASE COMPLETE THIS QUESTION!!!

 

[wrugoin]

This article was such a freaking waste of effort. Why? Because you don't compare any of the benchmarks to the alternative HDD. What good is display the time of a level load test if you don't compare it to something?

SSDs are expensive. I have an HDD. I'm a hard core gamer and I care about launch and level load times. Hmmm it took the reviewer 0:58 to load a level in Crysis 2... is that good? Who knows? Do I invest in an SSD based on that data? I still don't know

Seriously, there was no value to this article if you're considering replacing an HDD with an SSD for gaming.

 

[brisingamen]

the difference between a sata 2 disc drive and a low level ssd just isnt worth it in space sacrifice or money, the latest generation of ssd's is awesome and faster in just about every way, ive used one and its great but at a cost that just, . . amazing, everyone i show them too it stunned by the price, the worst sticker shock ive ever seen on a customers face.

 

[Guest]

SSD's is just storage. Are we really splitting hairs now on loading programs into RAM? Really people. Yo would would spend the extra money for SSD drives just to load a program a little faster. Yet spend hours playing a game?? Unless you have money to burn, I can think of a lot better improvements for gaming then a SSD drive.

 

[alyon]

Reasons SSD are nice for gaming.
1. Level/map load times
2. Alt+tabing with little delay
3. Transfering large files/ anti virus while gaming likes it nothing

in game benefits such as fps are not apparent unless there is MAJOR reads/writes going on with very large levels. Think like Arma ÎI big.