PDA

View Full Version : eVGA e-GeForce 6600 GT PCI-E


Pages : [1] 2 3

Clay
01-22-05, 01:24 AM
eVGA e-GeForce 6600 GT PCI-E (http://www.nvnews.net/reviews/evga_6600gt_sli/)

http://www.nvnews.net/reviews/evga_6600gt_sli/images/box.jpg

This is one of my shorter reviews, skipped a lot of the fluff due to time constraints. Plus, I wanted to try something different in presenting the results. I'm just sick of bar, line and silly string graphs. :D Also, I was really just focusing on the difference between SLI disabled and enabled. So, here you go. :)

MUYA
01-22-05, 01:25 AM
SLI R0x0rs !

Good stuff Clay! :D

zoomy942
01-22-05, 09:18 AM
that was an excellent review. i love the new webster's case :)

BrianG
01-22-05, 10:49 AM
Nice work, Clay.

Based on the reviews and editorials around the web, I may just have to put together an op-ed on ho we are losing sight of the real advantage of SLI. I may have a few questions and benchies for you to run, if you don't mind???

Regardless, I like the "case". I have been toying with the idea of building a true computer desk. That is a real desktop PC. Heh. (Man, that was a bad joke.)

Clay
01-22-05, 11:01 AM
Thanks guys. :cool:

@BrianG: Yeah, good idea. There are many misconceptions about SLI (some are understandable, some are not).

Sure, feel free to post any questions or PM me if you like. Any benches that you'd like me to run would need to be done by this coming Tuesday (1/25) though.

GlowStick
01-22-05, 01:05 PM
Great review, I espicaly like the %Decrease with turning AA/AF on, brilliant. Anyways, I feel its a good article and it trys to clear up all of the Misinformation and the misinformation that the orginal misinformation has spawned! ; D

SH64
01-22-05, 03:01 PM
WOW these D3 timedemo1 numbers are unbelievable .. literally :p , i get much less than that on my 6800GT@400/1100 , P4:3.0Ghz setup which is around 50fps (16x12,2/8)
it could be the CPU thats pushing the score like this ? .. anyway
any chance running HL2's VST & some of FarCry demos ??
i'd like to compare the results on my XT-PE system to these .

i'll attach a batch file for FC1.3 - Research demo @1600x1200,4xAA,8xAF & i like to see what score you can get in it (if you have the time to do it)
just change the ext to .bat after you download it .

Oh & can you send me the file/script or whatever for the percentage calculator ?? i was looking for one for a long time :D

Clay
01-22-05, 04:57 PM
The CPU difference could be part of it, also remember that I used the 71.24 drivers (I had problems with 66.93s in HL2 that I didn't have time to mess with resolving). The 71.24s offer better performance especially with SLI.

Sure, I can run the FarCry benchmark you attached. I'll post back here with the results in a day or two.

I just wrote that percentage calculator. You can probably find other examples on the web. You can just view the source code of that page and you'll have what you need. On second thought, I'll give it to you below so you don't have to weed through the source of the page.

This goes somewhere in your page, prefereably in the <head><SCRIPT LANGUAGE="JavaScript">
<!-- Begin
function pctIncrease(frmPct) {
var val1 = frmPct.val1.value;
var val2 = frmPct.val2.value;
var pct = ((val2 - val1) / val1 * 100);
var rlength = 1; // The number of decimal places to round to
return Math.round(pct*Math.pow(10,rlength))/Math.pow(10,rlength) + "%";

}
// End -->
</script>This is the souce code for the <form> that that houses the controls for the calculator. You can place this anywhere within the <body> of your page. If I'm stating the obvious to you, sorry, just no sure what you know or not about it.
<form name=frmPct>
<table border="0" cellpadding="8" cellspacing="0" width="460" bgcolor="#dddddd" style="border: 1px solid #000000" id="table3">
<tr span class="sectionHeader">
<td align="center" bgcolor="#005599">Percentage Increase/Decrease Calculator</td>
</tr>
<tr span class="tableTitle">
<td align="left">
<span class="articleText">
<div align="center">

<table border="2" bgcolor="#ffffff" width="295" id="table4" cellspacing="0" cellpadding="9" style="border-collapse: collapse" bordercolor="#808080">
<tr>
<td class="articleText" align="left">

From:
<input type=text name=val1 size=4 style="background-color: #FFFFCC; font-family:Verdana; font-size:10pt">&nbsp;
To: <input type=text name=val2 size=4 style="background-color: #FFFFCC; font-family:Verdana; font-size:10pt">
<input type=button value="Calculate" onClick="this.form.pctchange.value = pctIncrease(this.form);" style="font-family: Verdana; font-size: 12px; color: #003399">
<br>
The percentage change is:&nbsp;<input type=text name=pctchange value=" " size=6 style="font-family: Verdana; font-size: 10pt; border: 1px solid #FFFFFF; font-weight:bold; padding-top:2px">
</td>
</tr>
</table>

</div>
</span>
</td>
</tr>
</table>
</form>

-=Gib-McFragger=-
01-22-05, 06:15 PM
Yea, well my SLI Voodoo setup will pwn your system. :p

hehehehe Great review man and I like your 1337 new charts.

Erm...pardon me for ingnoring you at the end on MSN.........I fell asleep which did wonders for my oversleeping headache....... :barf:

SH64
01-22-05, 07:31 PM
If I'm stating the obvious to you, sorry, just no sure what you know or not about it.

Are you kidding ! without those details i wouldent know where the hell to put the scripts .. i'm a total n00b when it comes to html/java langs :o

& thanks a lot .. it worked ! :D

Clay
01-22-05, 09:32 PM
@Gib. That's fine, just ignore me. :D

@SH64: No problem, glad you got it working. Oh, and one tidbit of info. It's a common mistake that people make to refer to JavaScript as Java. They're actually very different and not really related at all. Netscape developed JavaScript (actually was called LiveScript initially). At any rate, Java is not used for that quick little calculator, just JavaScript. :)

Clay
01-22-05, 10:19 PM
CHANGE LOG UPDATE

January 22, 2005 - Clarified that the timedemos were run multiple times and an average was then taken to achieve final frames per second for the final result on page 2. Also clarified that anti-aliasing and anisotropic filtering settings were set to "Application Controlled" within the Forceware Driver Control Panel.

Kojiro
01-22-05, 10:22 PM
I believe that SLI's true purpose is for users who buy one 6600GT PCI-E now. Then to improve their graphics in the future, they install a second one. That way you can improve graphics without wasting the previous investment. In other words, you save money.

This is why

If an AGP 8x or non SLI capable motherbord user bought a single 6600GT, and later wished to get better graphics and go for a 6800GT, they would have to spend 400 on the 6800 GT and waste the 200 spent on the 660GT. The result is that user spent 600 dollars. SLI solution is only 400. And as new cards are being released, the price for the 6600GT drops. In other words, the longer you wait, the cheaper it is to SLI. Imagine the price cuts and rebates that will be availible probably by the end of this year and even beter ones next year.

This means even if you are not too extreame on graphics,and don't want to spend too much on a graphics card, then SLI will alow you to use just one card just like a normal PCI-E board, then when you feel like things seem too slugish, you can buy the same card again, but now at a lower cost do to price drops or rebates. Or the best of all price drop + store rebate + manufacture rebate. :D

the lonly thing holding SLI back is only Nvidia boards suport it. Correct me if I am wrong, but this limits SLI to AMD users. Intel despite playing catchup to AMD in terms of processor performance. Intel still dominates the majority of the PC processor market. Nvidia does not make Intel mother boards because they are not licenced to do so. But intel could incorporate SLI (they would have to colaberate with nvidia) into their chip set if they were not so highly backed by Microsoft which have been working hand in and with ATI. In other words, Intel does not need SLI mother boards since ATI (haven't you noticed that a lot of Intel based laptops have a mobile radeon on it? AMD based laptops most likely have a Geforce Go, unless it's not using nvida chipset.) will not suport it. The cost of a SLI board is quite expensive, one could save their money, buy their choice of low costing AGP motherboard, and get a high end card.

Thus if SLI does not catch on like it did in the past, then all that will be left is memories of a failed product, sad to say.

Clay
01-22-05, 11:42 PM
CHANGE LOG UPDATE
January 22, 2005 - Updated the Doom 3 benchmarks on page 2 with 1024x768 and 1280x1024 timedemo results to help illustrate the memory bandwidth restriction points.This goes along with what BrianG was talking about earlier, you can see these new results below as well.

http://www.nvnews.net/reviews/evga_6600gt_sli/images/doom3_trdelta1.gif

MikeC
01-23-05, 12:07 AM
Oh & can you send me the file/script or whatever for the percentage calculator ?? i was looking for one for a long time :D

Try this link: http://www.newsengin.com/neFreeTools.nsf/PercentChange/formPercentChange

MikeC
01-23-05, 12:13 AM
Correct me if I am wrong, but this limits SLI to AMD users. Intel despite playing catchup to AMD in terms of processor performance. Intel still dominates the majority of the PC processor market. Nvidia does not make Intel mother boards because they are not licenced to do so. But intel could incorporate SLI (they would have to colaberate with nvidia) into their chip set if they were not so highly backed by Microsoft which have been working hand in and with ATI.

This will change in the near future.

NOVEMBER 19, 2004 - NVIDIA and Intel Sign Broad Cross-License and Chipset License Agreements

http://www.nvidia.com/object/IO_17070.html

http://www.nvnews.net/cgi-bin/search_news.cgi?keyword=nforce5&firstform=1

Clay
01-23-05, 12:16 AM
Try this link: http://www.newsengin.com/neFreeTools.nsf/PercentChange/formPercentChange
Yeah that one is always in my online utils bookmarks. :cool:

Clay
01-23-05, 01:02 AM
WOW these D3 timedemo1 numbers are unbelievable .. literally :p , i get much less than that on my 6800GT@400/1100 , P4:3.0Ghz setup which is around 50fps (16x12,2/8)
it could be the CPU thats pushing the score like this ? .. anyway
any chance running HL2's VST & some of FarCry demos ??
i'd like to compare the results on my XT-PE system to these .

i'll attach a batch file for FC1.3 - Research demo @1600x1200,4xAA,8xAF & i like to see what score you can get in it (if you have the time to do it)
just change the ext to .bat after you download it .

Oh & can you send me the file/script or whatever for the percentage calculator ?? i was looking for one for a long time :D

Here are the results from the Research timedemo in FarCry based on the batch file you provided:


CORRECTION: RESULTS BELOW ARE @ 1024x768 0xAA/0xAF and low-to-medium video option settings


SLI Disabled
================================================== ============
TimeDemo Play Started , (Total Frames: 1512, Recorded Time: 46.65s)
!TimeDemo Run 0 Finished.
Play Time: 14.75s, Average FPS: 102.51
Min FPS: 80.69 at frame 90, Max FPS: 149.97 at frame 808
Average Tri/Sec: 2540808, Tri/Frame: 24784
Recorded/Played Tris ratio: 1.71
!TimeDemo Run 1 Finished.
Play Time: 13.43s, Average FPS: 112.61
Min FPS: 80.69 at frame 90, Max FPS: 154.59 at frame 754
Average Tri/Sec: 2801301, Tri/Frame: 24875
Recorded/Played Tris ratio: 1.71
TimeDemo Play Ended, (2 Runs Performed)
================================================== ============


SLI Enabled
================================================== ============
TimeDemo Play Started , (Total Frames: 1512, Recorded Time: 46.65s)
!TimeDemo Run 0 Finished.
Play Time: 13.13s, Average FPS: 115.14
Min FPS: 75.73 at frame 84, Max FPS: 158.39 at frame 887
Average Tri/Sec: 2850862, Tri/Frame: 24759
Recorded/Played Tris ratio: 1.72
!TimeDemo Run 1 Finished.
Play Time: 11.25s, Average FPS: 134.37
Min FPS: 75.73 at frame 84, Max FPS: 171.41 at frame 795
Average Tri/Sec: 3340281, Tri/Frame: 24858
Recorded/Played Tris ratio: 1.71
TimeDemo Play Ended, (2 Runs Performed)
================================================== ============

SH64
01-23-05, 02:35 AM
Hmmm something must be wrong Clay .. did you make sure these scores were @1600x1200,4xAA,8xAF ??

Clay
01-23-05, 02:40 AM
Hmmm something must be wrong Clay .. did you make sure these scores were @1600x1200,4xAA,8xAF ??I ran these pretty quickly so something could be wrong. However, what I did was the following: Clean install of FarCry Installed 1.3 patch Just ran your batch file (which specified 1600x1200, 4xAA and 8xAF) copied the results for research.log and posted them here.

SH64
01-23-05, 02:49 AM
I ran these pretty quickly so something could be wrong. However, what I did was the following: Clean install of FarCry Installed 1.3 patch Just ran your batch file (which specified 1600x1200, 4xAA and 8xAF) copied the results for research.log and posted them here.

of course .. i set the batch file to run @1600x1200,4xAA,8xAF however did you watch the demo while it was running & checked at least the res ?
because these numbers are way too high !! .. i get ~ 50fps on a single 6800GT , Jakup got ~ 80fps on a dual 6800Us .

Clay
01-23-05, 02:52 AM
of course .. i set the batch file to run @1600x1200,4xAA,8xAF however did you watch the demo while it was running & checked at least the res ?
because these numbers are way too high !! .. i get ~ 50fps on a single 6800GT , Jakup got ~ 80fps on a dual 6800Us .
It sure looked like 1600x1200 while it was running but I was doing about 10 other things while that was going on. :) I'll double check and report back here in this thread.

Clay
01-23-05, 03:08 AM
You're right. Those results I posted were @ 1024x768. The 1600x1200 4xAA/8xAF results are coming. I thought the command line params overrode the game settings, evidently not.

SH64
01-23-05, 03:16 AM
You're right. Those results I posted were @ 1024x768. The 1600x1200 4xAA/8xAF results are coming. I thought the command line params overrode the game settings, evidently not.

Strange ! they do override the settings for me .. maybe you were not in devmode ? ..

Clay
01-23-05, 03:19 AM
The -devmode flag is right there in the command line of your batch file...so I would assume so.

I didn't go with the auto-detect settings in FarCry. I set everything to "very high" (along with 4xAA and 8xAF of course). I'll post the results here in just a second.