Guild Wars 2

Hi,

my current load time is about two minutes. That is when I start the game and the game makes initial load to LA. Subsequent area changes are faster (half a minute).

Would an SSD help in faster loading?

Thanks!

Yes times infinity. My loading times to LA were similar (especially the first load after launch and during some world event, like Wintersday), I got an SSD a few days ago and now the first load is about 15 seconds.

The difference isn’t just noticeable, it’s ridiculous.
I have completely abandoned HDD’s in favor of SSD.
HHD’s with a stupid spinning disk, moving parts and gears, and heat are long obsolete.

SSD is vastly superior in every way except amount of storage vs price.

and with the program primo cache your SSD IS as fast as a ram disk.

I run 2 SSD’s primary is for the OS, and the other is dedicated to games,

You will never go back to a normal HDD after running an SSD its night and day,

Mapping into LA and zones that have LS content up where there are 100’s of people standing on a spawn point can still take upwards of 15-20seconds, but for average loading, your talking under 10 seconds including WvW.

Ok, you got me! I’m ordering one!

It’s a trap! don’t buy SSD!!

Just kidding I use 240 GB SSD for my games and the difference is amazing. Guys before me didn’t hype it for nothing. It’s great, especially in WvW where you’ll be able to move between waypoints in a matter of few seconds.

All MMO’s are textbook examples of software that benefits from ssd. Lots of reading small chunks of data across the drive, writing almost nothing. Go for it.

For some reason I was an exception to the rule.

I was using a SATA 3 HDD drive and recently upgraded to a Samsung 840 EVO SSD and installed my Windows OS GW2 on that instead. My load times were about 20-25 seconds for lions arch on the HDD and are about the same using the SSD… so…

The SSD is fine. Ran benchmarks and the speeds are good, so I guess I just had a really fast HDD lol. It also didn’t appear to make a major difference in how fast Battlefield 4 loads.

I personally didn’t see any benefit except that windows loads faster on startup, lol.

Sir Kaboomski.1508:

For some reason I was an exception to the rule.

I was using a SATA 3 HDD drive and recently upgraded to a Samsung 840 EVO SSD and installed my Windows OS GW2 on that instead. My load times were about 20-25 seconds for lions arch on the HDD and are about the same using the SSD… so…

The SSD is fine. Ran benchmarks and the speeds are good, so I guess I just had a really fast HDD lol. It also didn’t appear to make a major difference in how fast Battlefield 4 loads.

I personally didn’t see any benefit except that windows loads faster on startup, lol.

Then you must’ve had an amazing HDD. You still have that right? if you run the benchmark in Magician does it really come close to your 840 EVO?

As for the benefits, aside from the speed, there are no moving parts meaning less chance of a mechanical failure, also it doesn’t really heat up because of that and shaking doesn’t affect an SSD. Some HDD’s would freeze if you shook it, SSDs don’t have this problem.

Sir Kaboomski.1508:

For some reason I was an exception to the rule.

I was using a SATA 3 HDD drive and recently upgraded to a Samsung 840 EVO SSD and installed my Windows OS GW2 on that instead. My load times were about 20-25 seconds for lions arch on the HDD and are about the same using the SSD… so…

The SSD is fine. Ran benchmarks and the speeds are good, so I guess I just had a really fast HDD lol. It also didn’t appear to make a major difference in how fast Battlefield 4 loads.

use sata 2 instead and I promise you will see performance increase.
there are problems with sata 3. like my RIVE mobo for example sata3 is controlled by a
asmedia driver. and the driver is garbage. I have the same SSD, the Samsung 840 and I get
5Gb a second on sata2.

NeedCoffee.1402:

Sir Kaboomski.1508:

For some reason I was an exception to the rule.

I was using a SATA 3 HDD drive and recently upgraded to a Samsung 840 EVO SSD and installed my Windows OS GW2 on that instead. My load times were about 20-25 seconds for lions arch on the HDD and are about the same using the SSD… so…

The SSD is fine. Ran benchmarks and the speeds are good, so I guess I just had a really fast HDD lol. It also didn’t appear to make a major difference in how fast Battlefield 4 loads.

use sata 2 instead and I promise you will see performance increase.
there are problems with sata 3. like my RIVE mobo for example sata3 is controlled by a
asmedia driver. and the driver is garbage. I have the same SSD, the Samsung 840 and I get
5Gb a second on sata2.

Hrm, well that’s interesting.. I’ll try that when I get home from work. I have the same asmedia driver on my ASrock board.

NeedCoffee.1402:

I have the same SSD, the Samsung 840 and I get
5Gb a second on sata2.

Hate to be that guy, but I’ll just leave this here,
Difference between SATA I, SATA II and SATA III

SATA II (revision 2.x) interface, formally known as SATA 3Gb/s, is a second generation SATA interface running at 3.0 Gb/s. The bandwidth throughput, which is supported by the interface, is up to 300MB/s.

SATA III (revision 3.x) interface, formally known as SATA 6Gb/s, is a third generation SATA interface running at 6.0Gb/s. The bandwidth throughput, which is supported by the interface, is up to 600MB/s. This interface is backwards compatible with SATA 3 Gb/s interface.

You must live in the future to sustain 5Gb/s! I shall henceforth, call you Future Man!

Unless you test your SSD/HDD with IOMeter you do not know what performance metrics you are getting.

While SSDs will out peform HDDs 3:1 MOST OF THE TIME, there are instances where they are 1:1.

Load up IOMeter and run 4 Workers (1 per CPU, Ignore HT, if you have an 8 core CPU then run 8 workers). On each worker Build a test profile.

You will edit the test profile 8 times, once for each test.
64K 100% Read – Random
64K 100% Write – Random
4M 100% Read – Random
4M 100% Write – Random

64K 100% Read – Sequential
64K 100% Write – Sequential
4M 100% Read – Sequential
4M 100% Write – Sequential

On Top end SSDs you will see that while Random Reads/Writes will push the SSD to its max, Seq Reads/Writes bottom out at 4MB Block Sizes. My OCZ Vertex4 will push 350MB/s on 64K and 4M randoms, but Bottoms at 125MB on Sequential. My WD Black Spindles Will Bottom out at 35MB/s on Random Reads/Writes, but outperforms my SSDs on Sequential at 165MB/s

Then you have the 2 different types of SSDs to consider, SLC and MLCs.

MLCs are TERRIBLE, and found in 99% of the cheapest hdd sales. They provide some serious read ability, but have HORRIBLE writes. And they burn out the fastest, so make sure TRIM is enabled for MLCs. SLCs are the best and most expensive SSDs on the market. They are found in the 179+ Range.

And in this write up, im not even going over IOPS (operations per second) against the SSD/Disk, mainly because it does not play for this game. The max IO this game seems to push is about 80, and thats when you load into LA on a busy night. Most HHDs will push 75IOPS, max (some can push 85 if they are high quality and limit their platter density). So I am talking purely Read/Writes here.

And while your SataII,III,6 connections can push 3Gbps, 6Gbps…ect, You are limited at the Drive’s ability. To get your true SataII+ port speeds, you would need to stack 2-3 Drives per port and setup Raid0. So in reality, the difference between SataII+ is minimal in these cases.

didja want to see my crystal Mark benches?

NeedCoffee.1402:

didja want to see my crystal Mark benches?

Crystal mark is synthetic. Io-meter is not. So no.

sirsquishy.8531:

Unless you test your SSD/HDD with IOMeter you do not know what performance metrics you are getting.

While SSDs will out peform HDDs 3:1 MOST OF THE TIME, there are instances where they are 1:1.

Load up IOMeter and run 4 Workers (1 per CPU, Ignore HT, if you have an 8 core CPU then run 8 workers). On each worker Build a test profile.

You will edit the test profile 8 times, once for each test.
64K 100% Read – Random
64K 100% Write – Random
4M 100% Read – Random
4M 100% Write – Random

64K 100% Read – Sequential
64K 100% Write – Sequential
4M 100% Read – Sequential
4M 100% Write – Sequential

On Top end SSDs you will see that while Random Reads/Writes will push the SSD to its max, Seq Reads/Writes bottom out at 4MB Block Sizes. My OCZ Vertex4 will push 350MB/s on 64K and 4M randoms, but Bottoms at 125MB on Sequential. My WD Black Spindles Will Bottom out at 35MB/s on Random Reads/Writes, but outperforms my SSDs on Sequential at 165MB/s

Then you have the 2 different types of SSDs to consider, SLC and MLCs.

MLCs are TERRIBLE, and found in 99% of the cheapest hdd sales. They provide some serious read ability, but have HORRIBLE writes. And they burn out the fastest, so make sure TRIM is enabled for MLCs. SLCs are the best and most expensive SSDs on the market. They are found in the 179+ Range.

And in this write up, im not even going over IOPS (operations per second) against the SSD/Disk, mainly because it does not play for this game. The max IO this game seems to push is about 80, and thats when you load into LA on a busy night. Most HHDs will push 75IOPS, max (some can push 85 if they are high quality and limit their platter density). So I am talking purely Read/Writes here.

And while your SataII,III,6 connections can push 3Gbps, 6Gbps…ect, You are limited at the Drive’s ability. To get your true SataII+ port speeds, you would need to stack 2-3 Drives per port and setup Raid0. So in reality, the difference between SataII+ is minimal in these cases.

So am I supposed to use to 15 year old version? or the 8 year old version?
or better yet link me the ver. that would make you happy.

the year of the version doesn’t matter. its the same tool used to measure IOPS on 250k+ USD Sans (yes, I am also a storage Architect).

so i can use the one from the 80’s? sweet, that should be accurate.

can you explain in your best storage architect voice why crystal disk isn’t good enough?

NeedCoffee.1402:

so i can use the one from the 80’s? sweet, that should be accurate.

can you explain in your best storage architect voice why crystal disk isn’t good enough?

With that tone, no. even responding to this is a waste of my time now.

sirsquishy.8531:

NeedCoffee.1402:

so i can use the one from the 80’s? sweet, that should be accurate.

can you explain in your best storage architect voice why crystal disk isn’t good enough?

With that tone, no. even responding to this is a waste of my time now.

tone? like you or all the other negative nancies don’t have a tone?
Ohhhh but i poke fun at you antique software I all of a sudden a tone.

you people aren’t easy be friendly around.

Well until i learn how to use this overly complicated nonsense, my 5gb per second on crystal disk will have to suffice.

NeedCoffee.1402:

sirsquishy.8531:

NeedCoffee.1402:

so i can use the one from the 80’s? sweet, that should be accurate.

can you explain in your best storage architect voice why crystal disk isn’t good enough?

With that tone, no. even responding to this is a waste of my time now.

tone? like you or all the other negative nancies don’t have a tone?
Ohhhh but i poke fun at you antique software I all of a sudden a tone.

you people aren’t easy be friendly around.

Well until i learn how to use this overly complicated nonsense, my 5gb per second on crystal disk will have to suffice.

To be fair, your knocking software that is used by extremely high level Engineers to measure and gauge very expensive Storage hardware…cause its OLD.

Your 5Gbps would be about 619MB/s. Only the VERY expensive Intel SSDs are capable of that xfer rate. And even then, only when they are running a RAID setup.

Attached are some Screenshots between my SSD and my 3TB Spindles(Raid1)

As you can see, the Spindles take a huge hit on SEQ Writes, but are comparable to SSDs on Reads.

NeedCoffee.1402:

Well until i learn how to use this overly complicated nonsense, my 5gb per second on crystal disk will have to suffice.

You know what, I’ll bite. Run AS_SSD and show your 5Gb/s results for seq read/write.

Here is the kitten bench.

I think you may have a defunct port or sensor, lol. Those aren’t your actual speeds :/
Your actual speeds will be somewhere around 500MB/s~.

In fact, those speeds would not be possible on the SATA standard. You CAN achieve those speeds with a dimmdrive via PCI-E, but you have other issues lol. If that were a true reading, you broke ground with a revolutionary new SATA port technology along with building your own motherboard and portraying an SSD you built yourself that would be the fastest SATA SSD! And thus, I refer back to your name, Mr. Future Man. Quite frankly I don’t know what would be causing that .

just because it doesn’t make sense to you doesn’t mean its not possible.

theses are my actual benches, i can do the same on my cheap mushkin SSD’s.

I have mentioned on this forum 20 times how to do it, but no one listens to me.

~futrureman

NeedCoffee.1402:

just because it doesn’t make sense to you doesn’t mean its not possible.

theses are my actual benches, i can do the same on my cheap mushkin SSD’s.

I have mentioned on this forum 20 times how to do it, but no one listens to me.

~futrureman

Its quite simple, you do not understand. Your Benches are synthetic and wrong.

Again, grab IOMeter, and run 4 workers with 45000 mapped sectors. Change the Block size between 64K and 4M for your SEQ-Read and SEQ-Write tests, then again for RND-Read, RND-Write. And you’ll get your REAL Throughput.

Refer to my screenshots for proper testing data (those are my actual numbers from last night)

NeedCoffee.1402:

and with the program primo cache your SSD IS as fast as a ram disk.

Lol, yeah right!

If you seriously think so, then good for you. You do know what primocache is now, don’t you? Also, benching your ssd with primocache active and thinking “this is my ssd performance!”, well whatever makes you happy I guess.

See, with primocache running, you aren’t benching the ssd, but what primocache has cached in ram.

OGDeadHead.8326:

NeedCoffee.1402:

and with the program primo cache your SSD IS as fast as a ram disk.

Lol, yeah right!

If you seriously think so, then good for you. You do know what primocache is now, don’t you? Also, benching your ssd with primocache active and thinking “this is my ssd performance!”, well whatever makes you happy I guess.

See, with primocache running, you aren’t benching the ssd, but what primocache has cached in ram.

that would make sense why his ‘SSD Bench’ was running at 1GB/s.

NeedCofee – RAMDisks are nice and all, but useless when doing benchmarks. If I had known you had a RAMDisk I would have told you to disable the driver before running ANY benchmarks ;-) – Even I use a RAMDisk, 8GB carved out of my 32G. Its very nice to have for reads. But throws stats for raw disk/ssd testing out the window!

performance gain is performance gain. there is no point in benching naked.
those numbers aren’t any kind of placebo, they are real.
instead of learning all those iometer bools, and variables I’ll let the developer show you.
http://www.romexsoftware.com/en-us/fancy-cache/demo-benchmark-iometer-1-defer-write.html

NeedCoffee.1402:

just because it doesn’t make sense to you doesn’t mean its not possible.

theses are my actual benches, i can do the same on my cheap mushkin SSD’s.

I have mentioned on this forum 20 times how to do it, but no one listens to me.

~futrureman

Oooooh, I didn’t realize you were caching it… That makes any bench test pointless with that enabled. That makes sense now though, so your 500MB/s cached is closer to 9GB/s, but only in cache. Seriously, I saw that pic and was so confused lol. I thought you came from the future with new product wanting to be heralded as the next Bill Gates

Not to worry, you will forever be my Mr. Future Man!

sobe.4157:

NeedCoffee.1402:

just because it doesn’t make sense to you doesn’t mean its not possible.

theses are my actual benches, i can do the same on my cheap mushkin SSD’s.

I have mentioned on this forum 20 times how to do it, but no one listens to me.

~futrureman

Oooooh, I didn’t realize you were caching it… That makes any bench test pointless with that enabled. That makes sense now though, so your 500MB/s cached is closer to 9GB/s, but only in cache. Seriously, I saw that pic and was so confused lol. I thought you came from the future with new product wanting to be heralded as the next Bill Gates

Not to worry, you will forever be my Mr. Future Man!

lol, thanks :-)

I’m starting to think we come from 2 different world’s. benching just for pure stats alone like it has some kind of meaning or is some kind of sport is silly to me. I just want the fastest computer I can afford and its fun tweak stuff to make it faster.

NeedCoffee.1402:

performance gain is performance gain. there is no point in benching naked.
those numbers aren’t any kind of placebo, they are real.
instead of learning all those iometer bools, and variables I’ll let the developer show you.
http://www.romexsoftware.com/en-us/fancy-cache/demo-benchmark-iometer-1-defer-write.html

You dont understand the Tech. RamDISKs are Cache. Used mainly for Reads. Benching with Cache Enabled skews the numbers and shows NO REAL performance of what you are benching

If the bench only uses the Cache you are not even touching your HDD/SSD in the test, as it never leaves the RAM.

And the Topic here is GW2 on kitten, not GW2 On a RAM DISK.

*Edit, lol they Kittened SSD.

sirsquishy.8531:

NeedCoffee.1402:

performance gain is performance gain. there is no point in benching naked.
those numbers aren’t any kind of placebo, they are real.
instead of learning all those iometer bools, and variables I’ll let the developer show you.
http://www.romexsoftware.com/en-us/fancy-cache/demo-benchmark-iometer-1-defer-write.html

You dont understand the Tech. RamDISKs are Cache. Used mainly for Reads. Benching with Cache Enabled skews the numbers and shows NO REAL performance of what you are benching

If the bench only uses the Cache you are not even touching your HDD/SSD in the test, as it never leaves the RAM.

And the Topic here is GW2 on kitten, not GW2 On a RAM DISK.

*Edit, lol they Kittened SSD.

no, I understand. I’ve been using similar programs for years. what I’m saying is it doesn’t matter if I’m getting those speeds in cache or not. as long as I’m getting those speeds.

I think I tried caching to HDDs a long time ago and I didn’t get the same gains I got caching SSDs.

NeedCoffee.1402:

sirsquishy.8531:

NeedCoffee.1402:

performance gain is performance gain. there is no point in benching naked.
those numbers aren’t any kind of placebo, they are real.
instead of learning all those iometer bools, and variables I’ll let the developer show you.
http://www.romexsoftware.com/en-us/fancy-cache/demo-benchmark-iometer-1-defer-write.html

You dont understand the Tech. RamDISKs are Cache. Used mainly for Reads. Benching with Cache Enabled skews the numbers and shows NO REAL performance of what you are benching

If the bench only uses the Cache you are not even touching your HDD/SSD in the test, as it never leaves the RAM.

And the Topic here is GW2 on kitten, not GW2 On a RAM DISK.

*Edit, lol they Kittened SSD.

no, I understand. I’ve been using similar programs for years. what I’m saying is it doesn’t matter if I’m getting those speeds in cache or not. as long as I’m getting those speeds.

I think I tried caching to HDDs a long time ago and I didn’t get the same gains I got caching SSDs.

Caching to SSDs is going to be alot faster then Caching to HHDs. BUT! it all depends on what you are doing.

64K vs 4M, then Seq vs RND, and Writes vs Reads are all going to lead to differnt outcomes on different setups. As you can see on the screenshots I posted above, my Spindles perform about the same as an SSD on Seq Reads, but Writes its about 60% of what the SSD can do.

And adding virtual cache (USB 3.0, Ram, or Flash mPCIE-SSD Cache) will invalidate any bench marks you are running against the Physical Disk/SSD. And that is the point here.

Whether or not you have gains from the cache is irrelevant, I want to know what your SSD’s raw benches are using your testing tool.

Whether or not you have gains from the cache is irrelevant, I want to know what your SSD’s raw benches are using your testing tool.

yeah, no I get that too. what I don’t understand is why?

I mean what’s the point?

Is that an 840 EVO Samsung? They will use you system RAM as cache to bench reads that high.

Meehael.8240:

Hi,

my current load time is about two minutes. That is when I start the game and the game makes initial load to LA. Subsequent area changes are faster (half a minute).

Would an SSD help in faster loading?

Thanks!

Hi OP
I have ran GW2 from a Sata2 and Sata3 SSD as well as a Sata3 HDD and “drum roll”, yes the game loads much faster. Hope you invest in kitten, they are amazing and I will not go back

Behellagh.1468:

Is that an 840 EVO Samsung? They will use you system RAM as cache to bench reads that high.

it’s not the evo. it’s the older pro. I purchased it 2ish years ago.