X1700 vs. 7700?

Same question ( again...)

Which one is better? In the Meaker's Notebook Graphics Charts, 7700 is marked as a 'Mid-High end', while X1700 just as a 'Mid range' card. But X1700 got the 'Avivo' support. Does this mean something important? Does Avivo has a significant effect on the card performance in general, or just give an extra boost on video-related works?

 

AVIVO doesn't have any affect of the performance of the card. AVIVO includes things like HDR, TV Tuner options, and things like that. Some aren't available in notebooks, because there are connection difficulties. Performance of both cards is a little difference. Meaker has reted these cards correctly. The Go 7700 should perform slightly faster then the Mobility X1700, because of core advancements. The speed difference isn't much, it is similar to the speed difference in the Go 7600. However, the Mobility X1700 should be a cooler card, and it is still very powerful. If you are choosing between these two, in two different notebooks, the difference isn't something that should change your decidion. Consider things like RAM, processor and other things that matter to you first.

 

The 7700 actually leads the x1700 by a pretty good amount, especially when coupled with 512mb of DEDICATED video ram.

Even though it wont use all 512mb, it can use more than 256mb, which the x1700 cannot really do.

 

I found both of them in a same A8 specs. So, this is the only thing that I must consider... (beside the price different of course...) Likely, I have to buy the NB in Germany. Prices there are quite high tough. Before, I thought W7 was cheaper than A8, but now, a more-powerful A8 almost differ +/-300 Euros cheaper than a 7400-powered W7 there... I have to reconsider my choice again now...

 

the 7700 with the same dedicated memory as the x1700 is not faster. The x1600 was faster then the 7600 and there is no differenece with the upgrades.

stock drivers / overclock on x1700 (256mb dedicated / 256mb shared):


stock drivers/ overclock on 7700 (512mb dedicated):
Score (Overclocked) - 4743 (474 Core /
458 Memory)

As you can see both are relatively equal on stock drivers and max overclock. With omega drivers/latest nforce drivers ive heard reports of x1700 pushing 5000 and the 7700 breaking 5000 as well. this is 256mb dedicated vs 512dedicated.

As soon as reviews on the g2 come out with the x1700 512mb dedicated it will show that the x1700 > 7700. Currently people are basing their results on benchmarks between the two without the same dedicated memory. Although i will give the 7700 credit for added pipelines/all dedicated memory/HD-tv support.

 

Here you see why 3DMark05 is becoming an irrelevant benchmark. Sure, the x1700 can score close to the Go7700, but 3DMark05 a) doesn't take advantage of more than 256MB of video RAM and b)runs at 1024x768, which is a fairly low resolution these days and doesn't allow the effect of the 7700's added pixel pipes to be really seen. The old x1600 and Go7600 beat my GeForce Go6800 quite handily too, all at stock clocks, but try playing a game at 1440x900 and the Go6800 will beat the snot out of them. Why? More pipelines that just aren't used in 3DMark05.

 

Another thing about 3DMark06 is that it uses Shader Model 3.0, which is a better indication of how the cards will perform in more advanced games.

It's irrelevant for the new cards (GeForce 7, ATI X1000), yes. 3DMark05 is relevant for older cards like the ATI X300 - X800, because it was designed to stress cards of that era and it only uses up to Shader Model 2.0, which is what those cards have. The GeForce 6 is probably better suited to '05 as well, but they are well suited to '06 thanks to their SM 3.0 capabilities.

 

In 3dmark 06 your looking at the difference between 2300 (X1700) and 3000 (7700) points, that says a lot about which will last longest.

 

It's funny that you mentioned the memory size, but the thing is neither 3DMark05 or 3DMark06 takes advantage of more than 256MB of video memory. The textures of 3DMark05 were written to be within 128MB, while 3DMark06 was written with textures to be less than 256MB.

In most cases, resolutions of 1024x768 or 1280x800 can fit into 128MB memory sizes. 256MB's advantage comes through at resolutions like 1280x1024 and 1440x900. 512MB isn't really needed until 1680x1050 and 1600x1200, which I don't think the Go 7700 has enough power to run acceptibly. I don't think 256MB vs. 512MB memory size should be a determining factor between choosing the X1700 and the Go 7700. You're far more likely to run up against a memory bandwidth bottleneck before running into a memory size bottleneck.

In terms of comparisons between the X1600, Go 7600, and the Go 6800, I'm pretty sure that given the same systems and stock clocks the Go 6800 will score higher in 3DMark05. I'm pretty sure you're 3DMark05 score with the Go 6800 is higher than most X1600 and Go 7600 systems. It's just the rest of the system, CPU, RAM, etc., that is giving the X1600 and Go 7600 an advantage.

In terms of other differences between 3DMark05 and 3DMark06. It's true that none of the 3DMark05 tests are specifically written to use SM3.0, that doesn't necessarily give the advantage to SM2.0 cards. If you are using an SM3.0 card, 3DMark05 will run the test in SM3.0 mode. That doesn't make things harder for the SM3.0 card since no SM3.0 effects are used. What is used though is any SM3.0 speed enhancements such as effects that can be done in fewer passes compared to SM2.0. In theory, given the exact same hardware, it will get a higher 3DMark05 score running in SM3.0 mode than SM2.0 mode.

It often seems that 3DMark06 seems to favour nVidia cards (and 3DMark05 seems to favour ATI cards). For 3DMark06, part of the reason may well be that the nVidia cards are actually stronger since this is their 2nd generation SM3.0 card so they've had an opportunity to tweak and improve. The other thing though, may be the fact that nVidia cards support vertex texture fetch while ATI cards don't. 3DMark06 does use vertex texture fetch and the feature is defined for DX9.0c (although there is a loophole making it optional). There was a debate when the R5xx generation launched whether that lack of vertex texture fetch meant they don't meet DX9.0c certification and nVidia of course milked it for all it was worth. In any case, I don't believe vertex texture fetch is a very commonly used feature and ATI implements the same thing with render to vertex buffer anyways. I'm not sure how 3DMark06 implements vertex texture fetch (whether they use render to vertex buffer as an alternative) and how significant it's usage is.

In any case, at the end of the day the Go 7700 is faster than the X1700. The X1700/X1600 has always been limited by having only 4 TMUs anyways. In actual gaming performance, the Go 7700 will be faster, but it wouldn't be by a huge factor. Looking at the future, architecturally, ATI's pixel shaders are more advanced than nVidia's and have shown an advantage in theoretical tests using more complex SM2.0 and SM3.0 PS. ATI's programmable memory controller also offers more flexibility in continued driver-based software optimization. ATI and nVidia's vertex shaders are generally on par, although nVidia's VS seems to be noticably more efficient in SM3.0 dynamic branching, which was a design decision on ATI's part since they consider dynamic branching in the PS more important for the future and focused on that instead. Again, despite ATI's architectural advantages, any RV530 (X1600, etc.) or R580 (X1900, etc.) designs are limited by the 3:1 PS:TMU ratio.

 

Ok i agree with the above post now after considering what each 3dmark version recognizes and the factors considered.

 

the reason the x1700 in the g2 is as fast as the 7700 is that it is coming stock overclocked. Odds are very good it wont overclock more whereas the 7700 will...

Its quite probable the gpu is asus g2 only as well. Its not a really desirable machine to most people but for the people who do like it they might appreciate that.

My a8js scores are about 150% of the x1700 scores like the a8jp.

If you are in the market that sells both of these ok, pay at least 100-200 usd more for the 7700 I feel thats reasonable.

my overclocked 05 7700 score was 5164 btw. I hate arguing about 3dmark scores but why are you comparing based on joe schmoes score compared to larry schmoes score?
My overclocked 06 score is about 800 points higher than a x1700 score, the gaming performance between these two gpus is more than that its a large difference.

 

Well, i would say, it depends on the brand and model.
X1700 would outperform GF7700 if put in Acer 8210..

http://forum.notebookreview.com/showthread.php?t=93035
take a look at this..
"3dmark 2005
8204: 3300 (stock build), 4358 (fresh format)
8210: 4646 (stock build)"
This 3d05 scores even outperform Asus G1 (which proof that 8210 is better in SM2 games)

What if they put X1700 in it?
It would be a BEAST!

 

The 7600 is a very strong chip, the desktop GT version beats the desktop 6800U, I think the notebook GT version whacks the 6800 NU around. It has more shading power under the hood than the ATI chip. Now if they had based it off the x1650 chip with 24 PSUs then it would have been different....

 

youve got to concentrate a lot harder to understand the scores.

the 7700 in the g1 would have 05 scores of at least the same as mine, 5164

the 06 scores would again be just the same as mine, over 800 points more that the x1600

the x1700 is exactly the same as the x1600, except for the one in the asus g2. so an x1700 in an acer etc would have the same 05 and 06 scores.

 

naah..
Stock driver..

G1 scores
"3DMark2006: 2389
3DMark2005: 4247"
http://forum.notebookreview.com/showthread.php?p=1695618

8210 scores
"3dmark 2005
8204: 3300 (stock build), 4358 (fresh format)
8210: 4646 (stock build)"
http://forum.notebookreview.com/showthread.php?t=93035

"Stock driver..", no overclock

Also, there's no Acer X1700 GPU out there yet. So, nobody knows.
But my prediction is, if X1700 is put in 8210, it'll be a BEAST (just as the progression from 8204 to 8210). Nothing can beat it (just prediction).

 

Wow that Acer gets 4646 in 3DMark05 stock on a X1600? That's amazing. It's quite possible though since Acer stock clocks their X1600 at 472MHz/468MHz which is probably the highest stock speed for a X1600.

In terms of overclocked performance, it's not like a X1600 has never gotten higher than 5000 points in 3DMark05.

http://service.futuremark.com/compare?3dm05=1811219

The Acer 8204 gets 5174 in 3DMark05 when overclocked to 529MHz/558MHz. This is only using a Core Duo 2GHz T2500. Going to a Core 2 Duo and using the latest drivers like in the modern Acer 8210 can easily boost the score by a few hundred points.

Admittedly though, the X1600 doesn't do as well in 3DMark06. Still, a Core 2 Duo MacBook Pro onced overclocked to 520MHz/530Mhz gets 2563 in 3DMark06. I'm sure it's doing even better now since this was an early score done when the C2D MBP was just released to check whether the GPU was still underclocked and to see how far it can go.

http://forums.macrumors.com/showpost.php?p=3019543&postcount=2216

5174 in 3DMark05 (Core Duo) and 2564 in 3DMark06 (Core 2 Duo) were both scores achieved on an overclocked X1600 so you'd think the X1700 would provide some more headroom on top, especially if the strained silicon process allows more overclocking room.

 

I really think that x1700 and 7700 are on par though and the x1700 has better image quality (although im comparing x1600 vs 7600 there). although 3dmark shouldnt be what you base your purchase on.

 

I hope someone releases some fps results sometime soon

参考：Acer TM8204WLMi(X1600)Reference / Asus a8js

DOOM3

85.4 FPS/100.5 FPS

Half Life 2

136.42 FPS/151.57 FPS


unfortunately, the acer has a batter processor (7400) and I dont have the settings for the test but they dont look anti aliased and filtered and those will be the determiners in the final battle

 

Wow, so the 7700 is actually closer to it's bigger brother (7900GS) than it is to the 1600?

 

According to some benchmarks, even a Go 7600 will outperform x1700 in 3dMark06 enviroments. There are dozens of benchmarks out there, on different clock speeds and system settings, though. If you're an advanced user and are about to overclock your GPU to the maximum level of it, you probably know what you're doing and which GPU you really want.

If you're an average joe who just wants a good GPU, either is good, and you probably want to pick up the one that has the system around it nicer, or offers a better price. If it's still on par, then the 7700 will be a sound choice.

Average users benefit very little from speculating on how much you can cook your GfG7700 before the can blows off and your laptop is good enough to fry eggs on, just get the system that better suits your needs in general, and assume the two graphics cards are roughly equal. If it comes down to a choice between the cards, pick the 7700. Don't let the overclocking, factory stock clock speeds, and the millions of benchmarks get you too confused.

 

3dmark06 is a horrible benchmark, don't trust it as they kept changing the formula over the past few months

that said, i can't believe you bumped this thread which wasn't on the first page

you should have bumped this one at the bottom of the first page

7700 vs x1700 using real life benches (games)

the conclusion is that the cards take turns beating each other by 1-4 fps in games. so they are practically the same (so the extra $200 for the a8js isn't really worth when compared to the a8jp). it must be noted that the tester wanted to maximize eye candy. if you were to put most settings at low you might see a difference between the cards.

 

Like i said...

 

Let's use this thread to continue the conversation. Thanks guys,.