NotebookForums.com › Forums › Notebook Manufacturers › Acer Notebook Forums › Acer Notebook Reviews › Acer Aspire 5672 Review
New Posts  All Forums:Forum Nav:

Acer Aspire 5672 Review - Page 9

post #161 of 214
I' ve done some research and found out that the X1600 HM (Hyper Memory) models have a 64 bit memory interface, while the X1600 without HM , only dedicated memory , has a 128 bit memory interface. All the HM models have a 64bit instead of 128 bit .

The reason they reduced the memory interface could be cause of the shared memory usage, because video card manufacturers often use cheaper memory with a lower buswidth. HM is basically a marketing tool rather than a performance boost, since it gives the "right" to vendors to lie about the actuall dedicate memory of the chip.

Memory Interface is the bus inside the gfx card that connects the core (gpu) with the video ram of the card, having a 64 bit instead of 128 bit simply cuts half of the memory bandwith of the card

X1600 (128 bit) Memory Bandwith : 15.04 GB/s
X1600 HM (64 bit) Memory Bandwith: 7.52 GB/s

Having a low bandwith means the Core (gpu) is not fed , and thus cannot compute fast enough. This could explain how an X1600 can gain up to 100 frames in CS:S while the X1600 HM struggles to get playable frames.

You ll be suprised to know that a X1300 delivers a 10.4 GB/s that's 3 GB/s more than the X1600 HM, but they usually preform equall due to the incresead pipelines in the X1600 over the X1300 models.
post #162 of 214
[double post]
post #163 of 214
But lvcien has a hypermemory model and he said he could run games at full settings. I don't see anything on whether Filiab's is HM.

And if he doesn't have HM, wouldn't his 3DMarks score have been a lot higher than mine?
post #164 of 214
Quote:
And if he doesn't have HM, wouldn't his 3DMarks score have been a lot higher than mine?
Yes it would be a lot higher than yours, as i state above HM models have 64 bit memory interface , where the dedicated only memory models have 128 bit.
Quote:
But lvcien has a hypermemory model and he said he could run games at full settings. I don't see anything on whether Filiab's is HM
Maybe he hasn't, maybe he is lying (joke) The fact is that HM models are way slower compared to the non-HM versions of them.
post #165 of 214
Quote:
Yes it would be a lot higher than yours, as i state above HM models have 64 bit memory interface , where the dedicated only memory models have 128 bit.
Right, and mine was higher than his.
post #166 of 214
Quote:
Right, and mine was higher than his

Here is the site to back up everything i said so far

http://www.gpureview.com/show_cards....=366&card2=453

The synthetics test run by 3dMark dont always represent the actuall gaming performance. As you state before you are not able to play CS:S with playable frames.
My old laptop which had an ATI 9700 128 MB runs Half Life 2 smoothly with medium-high settings, 9700 is 2 generations older than the X1600, so the only reason for your low frames is the 64 bit memory interface, 9700 has a 128 bit memory intercafe.
post #167 of 214
I know 3DMarks doesn't reflect gaming performance. I normally would ignore it completely. However:

1. You stated if my card was 64 bit and Filiab's 128 bit, his would perform better and score much better in 3DMarks
2. Filiab's 3DMarks score was lower than mine.
3. Filiab's card performs better than mine in games.

Those three things don't add up. Do you see what I'm talking about?
post #168 of 214
I didnt know this X1600 had problems to get playable framerates on CS:S at all.

Being only 64bit would of course mean lower performance, but by no means can an x1300 be better than this x1600, because the core has much more power. if i remember correctly this X1600 can run CS:S with everything turned on and on high, on average between 70-100 fps.

The card may be 64 bit, but its still a strong performer. 64 bit and 128bit shouldnt cut performance in half, it should only reduce a certain %.
post #169 of 214
Thank you for understanding! Do you have any suggestions for how I could improve my performance?

I've been doing more testing with the latest Catalyst desktop drivers (the ones that gave me the 75% increase but were unstable). I figured out the performance increase is caused solely by Catalyst AI. Without Catalyst AI turned on, I get ~47fps in the Video Stress test in CS:S. With Catalyst AI on Normal I get 71fps and with it on Advanced I get 79fps. That's absolutely incredible. I'm going to do more testing.

In the Acer drivers, Catalyst AI doesn't affect game performance at all.

I'm going to play the game and tinker with the settings and see if I can get it to stay stable.

No, wait, 3DMarks first. I'll get back to you with my score in a few.
edit: 5072. Not much of an increase. Now for testing.

More info: CS:Source is still unstable with the latest drivers, and now I don't get any performance increase outside of the stress test. I've tried Chronicles of Riddick and SW: Knights of the Old Republic also. Riddick runs about the same or slightly better and has no stability issues. KotOR runs a lot better (only slows for smoke) and also appears to be free of stability issues.
post #170 of 214
Good God. Did a bench with ATT. Did worse than an X600 Mobile at 400/270.
post #171 of 214
This is weird indeed. I dont use the catalyst but the Omega Drivers. they should be the same tho.

My 64mb X700 with 3dmark 03 of 4400 and 3dmark05 of 1185, scores on that ATT test 1150, just above the 9600XT

Have you checked if your CPU is not overheating when playing games? that is the issue with my acer 4402.

Im not sure what could be causing you trouble running games like CS:S, considering you should perform better than me, because other 5672 users have reported positive performance with that game on the highest settings.

i would recommend checking if you are not overheating, and perhaps even use Omega drivers to see if it helps.

i always use high catalyst AI thing too.
post #172 of 214
This is so goddamn weird. I went back to the stable drivers I'd been using again and now the Catalyst AI is helping again, I get 71 with it on high.

I don't think it's a heat issue, since it doesn't gradually get slower or anything like that.

Earlier when I was using the latest Catalyst drivers and ATT, I used the on screen display to track the usage stats of the card as I played different games. It starts out with free video memory at 128 and free texture memory at around 370 (the numbers add up to 512, I'm not sure if the texture memory is just the shared memory). When I started a game in CS:S, the free memory kept dropping until the free video memory hit 0, when the game crashed. Remember, that was with unstable drivers.

In Riddick, the numbers always stayed at exactly 92mb video memory free and something like 302 mb texture memory free. Never twitched at all.

Then I tried KotOR (btw, the framerate indicated it wasn't running as much faster as I thought it was), and the numbers for free memory never dropped at all, they stayed at max. Not sure if that means it wasn't using my video card at all (would explain the performance issues) or ATT reported the info incorrectly.
post #173 of 214
edit: Ah, ****it, not sure I actually know what I'm talking about here.
post #174 of 214
The memory interface is what defines low-mid-high end cards, 64-128-256 bit, taking two similar cards that only different in the memory interface , lets say 64 bit over 128 bit they are going to produced different results, with the 128 bit model getting the higest performance.

No matter what drivers you ll use , no matter how much overclocking you make the bottleneck still remains and its a hardware issue that can't be resovled. If you are able to play todays game with reasonable frames enjoy cause you wont be able to play any future ones.

Enough said , this thread is going for too long.
post #175 of 214
Okay, you are obviously not understanding what I'm saying. Other people with exactly the same card as me are getting much better performance than I am. That's not right. It doesn't matter if it's 128 bit or 2 bit, other people with the 128/512mb HM Mobility X1600 and the Acer Aspire 5672 1.66ghz Intel Core Duo are getting much better game performance than me. Do you understand the problem? I don't care how crappy the card is, only that it's doing much worse than everyone else's.
post #176 of 214
I was wondering if you had any artifacts in your games. My acer had a bad video card that would infrequently artifact and underperformed. Could be bad video memory. I would contact acer and let them diagnose it, they may just throw a new mobo in it under warranty. Thats what they did with mine within a 10 day turn around. Sorry to hear you're having trouble, I've been there, its not fun

Cheers
post #177 of 214

my x700

Also, I have the TM 4404 with a Turion ML37 and a 64mb x700 which is rumored to have a 64bit memory interface and I can play every current game I've ever loaded. Also, I score a 1700 on the ATT benchmark so I definitely think something hardware oriented is wrong with ur video card. Question, has it been proven that the 128mb x1600 with HM is only a 64bit? I find that hard to believe.
post #178 of 214
I get serious artifacts in Riddick but nothing else.

Every resource I've looked at says the HM is only 64 bit. Granted, that's not a lot of resources. I'm certainly no expert, half of the stuff I know about video cards I just learned trying to solve this problem I'm having. I didn't even know HM was bad (or that my card had it) until this.

My lappy is still under warranty but I need it for school, can't send it in until winter break in December. Also, you have to sign a waiver letting them replace your hard drive if you feel it is necessary, but I absolutely refuse to have my hard drive replaced (although it shouldn't be necessary, as it has no problems). If it's not too much I'll buy another drive and swap it in (assuming it is in fact a bad card).

Thanks, and to everyone else. Sorry I forgot to say thanks sooner. If anyone else has any suggestions I'm still open.
post #179 of 214
you don't have to sign that waiver, I didn't since my HD had been upgraded to a 100gb 7200rpm from the crappy 120gb 4200rpm. But ya if you need it for school I understand that. I still don't believe there is a 64bit memory interface, if that were true, the laptop wouldn't score well at all. Sounds like your laptop is defective. What is your 3dmark05 score, my lappy gets a 1420. If you don't like 3dmark (i don't) Aquamark 3 tends to be more video oriented and more indicative of real game performance. I score a 40K on that one where my desktop (A64 with 6800 ultra) gets a 70K which is pretty consistent with my game performance.
post #180 of 214
I got 36,000ish in AquaMark just now, with the standard settings. I haven't tested using the laptop's native res yet.

I got a ton of artifacts in AquaMark's menu.

edit: Oh, you need the professional version to run with adjusted settings. I couldn't tell because of the artifacts. Look at this, this doesn't happen in anyhting else. http://img.photobucket.com/albums/v2...quamarkwtf.jpg
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Acer Notebook Reviews
NotebookForums.com › Forums › Notebook Manufacturers › Acer Notebook Forums › Acer Notebook Reviews › Acer Aspire 5672 Review