Sign in with
Sign up | Sign in
Your question

Zotac launches its 9800GT Eco card

Last response: in Graphics & Displays
Share
March 17, 2009 9:26:48 PM

http://www.fudzilla.com/index.php?option=com_content&ta...

We're getting closer to the 8800 GTX speeds with out a power connector as each day passes, this is the only card that looks like it can beat the 4670 till now with out having a power connector.

Very interesting

Enjoy!
a b U Graphics card
March 17, 2009 10:23:50 PM

I'd have to see benchmarks to believe it's faster than a 4670.

'lower clock speeds' on a 9800, doesn't that make it a 9600gsomething? :D 
March 17, 2009 10:42:52 PM

no

112 SP(9800 GT) vs 64 SP (9600 GT), Def not 9600 GT. And even if it were 9600 GT, it would still be stronger than the 4670, only prob is how much?

"The new 9800GT Eco works at 550MHz for the core, which is 50MHz lower than the reference 9800GT card. It still has 112 stream processors but the shader clock has been dropped from 1,500 on the reference card to 1,375MHz for the Eco Edition. The card has 512MB of GDDR3 memory clocked at 1,800MHz and a 256-bit memory interface."

Only 50 mhz decrease in core clock, but looks like the shader went to 1350 compared to 1500.

I wonder how much, if any, OC potential this card has.

There is not if, this card will beat the 4670, look at the specs, whole different league from the 4670. The lower clocks don't even look like they will drag it to 9600 GT levels. It will rest between 9600 GT and 9800 GT.
Related resources
a b U Graphics card
March 17, 2009 10:56:13 PM

It's a nice looking little card tbh. If what it claims is actually true, it would have made a nice 240. At least it seems to be a real improvement compared to the 250.
March 17, 2009 10:56:23 PM

There is no doubt it will be faster than the 4670, but another thing that makes the 4670 great is the price. Its so cheap for so much performance.
Will this be the same?

Also, would you even be able to overclock this any (since it IS a underclocked 9800gt)?? I mean, wouldnt overclocking make this card need the extra power again?
March 17, 2009 11:00:43 PM

^thats exactly what I was thinking, but who knows, I mean cards have managed to Overclock before with out power conncetors, but I know some cards are really picky.

example for my old 9800 GX2, there were 2 connectors, 6 pin and 8. You could connect 2 6 pins, but you couldn't overclock.

So really we will have to see.

Yes the price might be an issue, but I see this card going for 100$ or so. Hopefully less.
a b U Graphics card
March 17, 2009 11:02:50 PM

Thing is, the 4670 isn't that far off the 9600gt in performance. We are talking <5fps in most games. The 9800gt isn't that much far ahead of the 9600gt either.

If you drop those clocks all around, you could easily see a 9800gt drop below a 9600gt in benchmarks. If it drops far enough...

http://www.hexus.net/content/item.php?item=15393&page=1...

That was a Zotac 9600gt at 650 clock, 1625 shader and 1800 memory. Compare that to this new 'energy efficient' 9800gt @ 550 clock, 1375 shader and you'll soon figure why it doesn't need extra power.

This '9800gt' is quite a bit lower specced than that 9600gt in this review - so much lower that I would have to say there is a good chance this new 9800gt would perform worse than their 9600gt, and possibly worse than the 4670 also.
March 17, 2009 11:38:52 PM

Grid is an ATI game:) 

And no, not as easily as you'd, remember it still has almost double the SPs, and almost double the texture units.

Look at the clocks, 50 mhz down, 125 shader and the memory is still the same. And what smore beatiful, it doesn't need a power connector.

The 9600 GT is generall 10-15% faster than the 4670 and from this chart its about 13%.



thus makes the 9800 GT in general 20-25%

From the specs, it won't loose that much:) 
March 17, 2009 11:48:06 PM

@jennyh
WHAT??
Are you out of your mind?? There is no way that even an underclocked 9800gt will perform worse than a 4670, absolutely no way.

There is a difference in the amount of shaders (9800gt has 112, 9600gt has 64), thats what makes the GPU a 9800gt anyway. Just because the clocks are lower doesnt mean the card is worse.
They didnt make the same chip for all the 9XXX series cards and just give them different clock speeds and coolers, it just doesnt work that way.

Also, keep in mind that a 9600gt still needs an extra power connector, this one doesn't.
a b U Graphics card
March 17, 2009 11:56:49 PM

All I'm trying to say is, I'd be a bit wary of this '9800gt'.

It has a slightly lower clock and its shader clock has been slaughtered to make it 'energy efficient'.

You don't simply go from a power hungry 8800 to a pci-e bus powered card without losing something. Not even in 2-3 years.

This card will perform worse than a 9600gt. If double shaders can only give a small 10% increase at stock, losing 50mhz core and 20% shader clock will place it lower in terms of raw performance. Or more likely it will be better in some games and worse in others.

I do believe it will be faster than a 4670, but I bet you will be surprised at just how small the gap is. I don't believe it will slot inbetween a 9600gt and 9800gt stock in price or performance. It will probably be priced the same as a normal 9800gt while performing slightly worse than a 9600gt.
March 18, 2009 12:06:57 AM

Im gonna have to agree with jenny on this one. This card has probably suffered on the performance more than we think. I mean the 6pin is removed for hells sake, that's 75W of power gone, and you lost it somewhere. This is a regular old G92, not some new efficient revision.
March 18, 2009 12:10:44 AM

Where is this power hungry card you speak of? The 9800 GT???? Really?

I am very doubtful that the card will be any less than the 9600 GT. I mean to shrink 15% to match the 9600 GT, doesn't seem possible.

20% shader, 10% core, well cards don't really work that way, its not like regular math.

Look at this, an Oced 9800 GT (1 gig I kno, but I'm looking at the low res), is 12% faster with 100 mhz OC, 243 mhz shader and 138 mhz memory increase (which is basically doubled)...it gains 12% generally, so that means that the downclock would only loose 6% by these numbers....

http://www.guru3d.com/article/galaxy-geforce-9800-gt-10...

so lets see, that bring it still over the 9600 GT:) 

We'll see when it comes out:) 
March 18, 2009 12:10:49 AM

If this card has higher frame rates than the 4670, and is ~$100, I would see people buying it. The reason most people buy the 4670 is because of the amount of power that it needs to run, but now another competitor will make it the power efficiency market more interesting.

Of course this isnt going to be able to run Crysis, heck no. Never expect that, just expect that it will be better performance than the 4670, which is the only card that offers great performance for such little power cost.
March 18, 2009 12:14:24 AM

Don't forget that this means that we can have sli setups in that strength area as well. not every1 has Crossfire boards, so now Nvidia holders will be able to sli their own 4670 card in a way;)
March 18, 2009 12:14:34 AM

L1qu1d said:
Where is this power hungry card you speak of? The 9800 GT???? Really?

I am very doubtful that the card will be any less than the 9600 GT. I mean to shrink 15% to match the 9600 GT, doesn't seem possible.

20% shader, 10% core, well cards don't really work that way, its not like regular math.

Look at this, an Oced 9800 GT (1 gig I kno, but I'm looking at the low res), is 12% faster with 100 mhz OC, 243 mhz shader and 138 mhz memory increase (which is basically doubled)...it gains 12% generally, so that means that the downclock would only loose 6% by these numbers....

http://www.guru3d.com/article/galaxy-geforce-9800-gt-10...

so lets see, that bring it still over the 9600 GT:) 

We'll see when it comes out:) 


I dont think its gonna drop down to the 9600GT, thats a little to far. But I do think it will being a larger drop than assumed.
March 18, 2009 12:18:33 AM

I'd say tops 10% if any, and I still thinkg 5-6% loss.

Remember these cards aren't energy hungry at all, event he 9600 GT isn't, they just aren't low enough not to require a 6 pin.

Only problem with this card is no OC potential, then again if you buy this card usually you know what your getting into.
a b U Graphics card
March 18, 2009 12:19:51 AM

I like this card. I'd love to see more stuff like this tbh.

I'm just very suspicious of something carrying the '8/9800' name now. Benchmarks will show this card to be less powerful that the 9600gt, of that I'm quite sure. If Zotac had done a 9600gt 'energy efficient' and not requiring more power then it would have been much more believable. The 9600gt went passively cooled, low profile etc - an energy efficient 9600gt card should have shown up before a 9800gt did...
a b U Graphics card
March 18, 2009 12:28:39 AM

The OC potential shouldn't be an issue. These things are rated on TDP, and a pci-e bus should be able to handle 75W in the worst case. The 4670 for example maxes out at around 60w or so, no matter how hard it is overclocked it will not require an extra power connector.

This zotac 9800 must be comfortably inside the 75w allowance of the pci-e bus even at the worst case. That extra power effeciency doesn't come free. Surely the 9600gt would have been much more likely to be pci-e powered?

I'm just very suspicious of the naming of the card. It might have the same number of SP's but it has 10-20% lower clocks than a 9600gt. Does that still make it a 9800gt? That's what I'm trying to say.
March 18, 2009 12:35:50 AM

Its more like a 9700M GT
March 18, 2009 12:38:31 AM

9800 GT alone has lower clocks than the 9600 GT?

Whats your point?

9600 GT has higher clocks than the 280 GTX lol Core wise haha...

Remember the 9600 GT doesn't run on the G92, it runs on G94 while the 9800 GT, even the underclocked, runs on G92.

G94 was created as a budget series. My laptop 9700M GTS runs on a G94 with 48 SP instead of 64, I need to OC it to 700 mhz for it to match the 9600 GT.
March 18, 2009 12:39:53 AM

spathotan said:
Its more like a 9700M GT


9700M Gt runs 32 SP, and is 8600 GTS speeds (desktop wise).

a b U Graphics card
March 18, 2009 12:50:43 AM

I just sold a 9800gt with 670 core, 1950 mem and 1675 shader clocks.

If anything, these 'energy efficient' 9800's will be failed g80's much like gainwards supposed 4850 with ddr5 memory.

The clocks are very low compared to even stock 9800's. I have no doubt they are the worst of the worst g80's rebranded as 'energy efficient'. I'm still quite impressed that they can run without an extra power connector, but don't be fooled as to what you are actually getting here. I think we are all cynical enough to realise what this card really is. :D 
March 18, 2009 12:59:12 AM

What the hell are you talking about it?

you completely lost me...G80? like the 8800 GTX, 8800 Ultras 8800 GTS?

Are we on the same page? Do you know what a revision is?

What are you talking about fooled? Your too confusing, especially with your ""

So your trying to say that they aren't energy efficient? maybe you haven't heard of the 2900 XT and most of ATi's high end cards?

I don't get what your argument is?

9800s failed G80s? that doesn't even make sense, seeing as they still sell, are realitivaly cheap, and are just rebranded 8800s GT.

Lets leave this as, we'll see when we get the price of the card:) 
March 18, 2009 1:01:23 AM

Alright so here is a thought: Did you ever think that they purposely used a 9800gt just so that it would get the same performance as a 9600gt, just without extra power? That makes sense, right? They are just trying to have a competitor with the 4670, simple as that. So no doubt it will be close to the 9600gt in performance, but the no extra power plug is to appeal to consumers who dont have a large power supply. I highly doubt that these are defective, though.
March 18, 2009 1:04:12 AM

Even if they were G80's (which they arent), I doubt anybody has held on to ones this long that were damaged bad enough to run on less than 75W at downclocked speeds.
a b U Graphics card
March 18, 2009 1:15:36 AM

Sorry I didn't mean g80's, what I meant was the older 65nm 9800gt's. It's not easy keeping track of which nvidia is which you know. :D 

What has the 2900xt got to do with anyway? As for my argument, thats pretty simple.

1) Where is the EE 9600gt? Don't you think that would be around before a EE 9800gt?

2) At what point does a 9800gt cease to become a 9800gt? If one of ATI's partners released a card with ddr5 and 50mhz core based on the r700, is it still a 4870? Energy efficient? I bet it would be.
March 18, 2009 1:16:46 AM

I could say the same about your arguments;) :p 
March 18, 2009 1:19:56 AM

jennyh said:

What has the 2900xt got to do with anyway?


References of energy efficency. The 2900XT is about the most inneficient piece of PC hardware ever created. And this new 4890 is looking like it might match it.
March 18, 2009 1:24:17 AM

thats completely different though, this is downclock and the 4850 with GDDR5 is a 4870 basically, a 9800 GT downclocked isn't architecturally a 9600 GT...

March 18, 2009 1:28:38 AM

Like I said I agree with you that it wont be as weak as a 9600GT. But its for sure going to come in just slightly above it. At that point its all about price, and your power supply. These would make pretty good moderate upgrades for people who weak PSU's, or dell systems and the like with PSU's you cant trust/dont want to bother with.

No power plug = any system.
a b U Graphics card
March 18, 2009 1:29:15 AM

Ok tell me this.

Why haven't we seen a 9600gt EE that didn't need a power connector. Looking at the charts, a reference 9600 GT pulls 98W compared to 105/125 of a 9800gt. Only last week a passively cooled 9800gt was released, a passively cooled 9600gt has been around for ages. It's all linked, power and heat.

It doesn't take a genius to figure out that Zotac have been smart here. They have been allowed to keep the 9800gt name even though it performs worse than a 9600gt. Double shaders running at lower clocks might allow them to keep the 9800gt name, but it won't let them keep the 9800gt performance.
a b U Graphics card
March 18, 2009 1:31:21 AM

L1qu1d said:
thats completely different though, this is downclock and the 4850 with GDDR5 is a 4870 basically, a 9800 GT downclocked isn't architecturally a 9600 GT...


Yes I know, but a 9600gt performs much better than it's architecture would seem to allow. That must have more to do with the higher clock speed than the number of sp's, don't you agree?
March 18, 2009 1:34:09 AM

jennyh said:
Ok tell me this.

Why haven't we seen a 9600gt EE that didn't need a power connector. Looking at the charts, a reference 9600 GT pulls 98W compared to 105/125 of a 9800gt. Only last week a passively cooled 9800gt was released, a passively cooled 9600gt has been around for ages. It's all linked, power and heat.

It doesn't take a genius to figure out that Zotac have been smart here. They have been allowed to keep the 9800gt name even though it performs worse than a 9600gt. Double shaders running at lower clocks might allow them to keep the 9800gt name, but it won't let them keep the 9800gt performance.


The archetecture differences and higher shaders alone will have it beat a 9600GT. But just how much is all up to how the clock speeds scale.

Perhaps somebody here with a 9800GT/8800GT can downclock their card to the Zotac Eco specs and run benches?
March 18, 2009 1:41:16 AM

jennyh said:
Yes I know, but a 9600gt performs much better than it's architecture would seem to allow. That must have more to do with the higher clock speed than the number of sp's, don't you agree?



Yes but along will come a game like Grid that loves SP. It will do well in games that don't demand the Sps and Text Units.

Its like my card, I need to OC it alot to match a 9600 GT, so overclocking can compensate for the SPs in certain cases, but you'll see games where that won't be enough.

there is no set rules for this. Don't forget that the 9600 GT came after the 9800 GT technically speaking:) 
a b U Graphics card
March 18, 2009 1:42:30 AM

It might do. Losing 20% shader clock is a lot though, even with double shaders. If all those shaders aren't getting used to the full, the 9600gt will be better due to shader/clock ratio.

Something has happened to turn a 105w TDP gpu into a 75w TDP gpu. It wasn't a die shrink or architectural improvements. Maybe the vcore has been dropped as well, that would fit in with energy efficiency at least and would make sense with a lower clock.

We will find out once the benchmarks are released.
March 18, 2009 1:48:21 AM

It doesn't make sense, then they could;ve just made a 9600 GT version of this card if this card performs worse than the 9600 GT.....It wouldn't make sense to do it with the 9800 GT if its going to perform sub 9600 GT.

No logic.

Shader clock for the GT will do nothing, this card from the specs will loose 6%, thats still 5-9% higher than the 9600 GT.

They would've been better off changing the 9600 GT, if this card fails to meet at least that...and trust me its not like we caught it, and their engineers didn't

So no this card won't be weaker than the 9600 GT.

No arguement
No logic
No go

9600 GT already consumes less than the 9800 GT, performs like a 9600 GT atleast, do you see the logic?
a b U Graphics card
March 18, 2009 2:13:15 AM

I do see the logic. It's all about selling the worst 9800 gpu's as energy efficient. There's nothing wrong with that except the usual which is many people will see the n800 tag and assume it's a better card than it really is, again.
March 18, 2009 2:22:12 AM

Econo will be post, ppl can't read or research well ;)  good luck:D 
a c 358 U Graphics card
March 18, 2009 3:13:46 AM

I would say Zotac's 9800GT Eco will be based on the G92b chip since it is less expensive for nVidia to produce than the G92 chip and due to the smaller die process it will also be more energy efficient. MSI's OC'ed 9800GT G92 video card has been measured to consume up to 83.2w, thus a down clocked 9800GT based on the G92b chip should be within the 75w envelop the PCI-e x16 port can deliver. Also since the GPU is down clocked, the vcore can also be undervolted as well which further reduces power consumption.

http://www.xbitlabs.com/articles/video/display/msi-9800...


Regarding the debate that this "Eco" 9800GT will be less powerful than a 9600GT, I would say it's doubtful because the GPU architecture is slightly different. See specs as follows:

http://www.xbitlabs.com/misc/picture/?src=/images/video...


By default, the clock speed of the 9600GT is higher than that of the 9800GT, yet we all know the 9800GT is the more powerful card. Dropping the GPU's speed by 50MHz is a 8.33% decrease, dropping the shader speed from 1500MHz to 1350MHz represents a 10% decrease. Let's simply assume there will be a 10% drop in performance (90% of normal 9800GT).

The following is a review of Zotac's overclocked 9800GT AMP, but it does include benchmarks of stock speed 9800GT and 9600GT. I will simply focus on the 1920 x 1200 benchmarks which are the most stressful.

http://www.techpowerup.com/reviews/Zotac/GeForce_9800_G...

Call of Duty 4
9600GT = 36.7 FPS
9800GT = 44.2 FPS

9800GT (@ 90%) = 39.8 FPS

Call of Juarez
9600GT = 42.2 FPS
9800GT = 57.3 FPS

9800GT (@ 90%) = 51.6 FPS

Company Of Heroes
9600GT = 107.1 FPS
9800GT = 143.8 FPS

9800GT (@ 90%) = 129.4 FPS

Crysis
9600GT = 15.2 FPS
9800GT = 18.9 FPS

9800GT (@ 90%) = 17.0 FPS

Enemy Territory: Quake Wars
9600GT = 46.6 FPS
9800GT = 54.5 FPS

9800GT (@ 90%) = 49.1 FPS

Far Cry - I liked this game, but come on it's kinda old.
9600GT = 117.6 FPS
9800GT = 130.3 FPS

9800GT (@ 90%) = 117.3 FPS

FEAR
9600GT = 54 FPS
9800GT = 60 FPS

9800GT (@ 90%) = 54 FPS

Prey
9600GT = 59.2 FPS
9800GT = 68.7 FPS

9800GT (@ 90%) = 61.8 FPS

Quake 4 - Ummm.... Is there something wrong with their benchmarks? The 9800GT is actually slower than the 9600GT?
9600GT = 30.2 FPS
9800GT = 18.7 FPS

9800GT (@ 90%) = 16.8 FPS

Splinter Cell 3: Chaos Theory
9600GT = 59.8 FPS
9800GT = 65.1 FPS

9800GT (@ 90%) = 58.6 FPS

S.T.A.L.K.E.R.
9600GT = 38.2 FPS
9800GT = 41.5 FPS

9800GT (@ 90%) = 37.4 FPS

Supreme Commander
9600GT = 36.3 FPS
9800GT = 42.5 FPS

9800GT (@ 90%) = 38.3 FPS

Team Fortress 2
9600GT = 58.4 FPS
9800GT = 66.8 FPS

9800GT (@ 90%) = 60.1 FPS

Unreal Tournament 3
9600GT = 62.4 FPS
9800GT = 71.9 FPS

9800GT (@ 90%) = 64.7 FPS

World In Conflict
9600GT = 27.0 FPS
9800GT = 32.0 FPS

9800GT (@ 90%) = 28.8 FPS


Conclusion

#1 The Zotac should perform between the 9600GT and 9800GT.

#2 Stop making half-ass guesses. There are plenty of benchmarks out there that can be used to make an educated "guess-timate" to what kind of performance to expect rather than simply saying a 9600GT will be faster than the "Eco".


You have been "edu-micated" go forth and prosper.
March 18, 2009 3:38:04 AM

@ the end of jaguar's post LOL

Ok, for 1, this isnt a conspiracy where they are trying to pass off bad chips. It just isn't. Im going to have to repeat myself now though....THEY USED THE 9800GT BECAUSE IT WILL GIVE THEM THE BEST PERFORMANCE WHILE BEING WITHIN THE REQUIRED POWER USAGE. If they used a 9600GT it would be closer to the 9500GT performance, and no one wants that. If they were selling the worse GPUs then how exactly would the GPUs be energy efficient? How would you make a defective GPU energy efficient?
March 18, 2009 3:57:46 AM

So i'm guess I was right:) 

Good to see back up facts 2:) 

Thank you jaguarskx
March 18, 2009 4:38:32 AM

My bet, as far as why we're seeing a 9800GT under 75watts and not a 9600GT, is that they tried both, and found that the 9800GT, even downclocked, was more powerful than the 9600GT, and was still able to fit in the 75watt envelope. They're trying to get an edge in performance over the HD 4670, while matching it's best factor (high performance : energy ratio).
a b U Graphics card
March 18, 2009 10:40:40 AM

Jaguar, why not show us the benchmarks at lower resolutions.

Nobody is going to be buying an energy efficient 9800gt and playing at 1900x1200, and the 9600gt dies at that resolution.
March 18, 2009 11:09:19 AM

So how come the 9800 doesn't when underclocked, right there its proven your so called power.

Lower resolution or not, its more powerful.

Thats what was trying to be proven, and it was.

Its not like lowering the resolution for these category cards will actually help one of the other, they both have 512 megs on a 256 bit board. Only difference is the SP and Texture Units (also clocks).

Your arguments are not logically acceptable, I don't even know how this can go on. I can see that if they made this card weaker than the 9600 GT, it would be pointless and they could've just used the 9600 GT.

What we are smarter then their engineers? oh my I better go out and grab a job, I don't need my diploma anymore:D 

I'm sure jaguar isn't done with you;).

a b U Graphics card
March 18, 2009 11:41:18 AM

I don't consider simply taking this eco 9800gt at 10% less performance proves anything. Anyone can pull numbers out of their behinds.

Why assume a 10% drop in performance? Let's take the previous TDP of 105w and compare it to the current TDP of 75w. That's a 30% reduction in power for only 10% reduction in performance? If Jaguar can do that, he should be working for Nvidia.

What if the number is closer to 20% instead?
March 18, 2009 12:14:50 PM

it wont be:) 

They would've grabbed the 9600 GT, with half the SP, and textures, and clocked that lower if anything, and would've saved ALOT more power. Considering its a 9% lower clock, and 9% lower shader, what do you think?

Memory speeds are the same. Everything is the same, when it comes to the card. POwer was reduced, and 9% clocks, how does that account for 20%?

Simple, it doesn't

a c 358 U Graphics card
March 18, 2009 10:55:05 PM

jennyh said:
Jaguar, why not show us the benchmarks at lower resolutions.

Nobody is going to be buying an energy efficient 9800gt and playing at 1900x1200, and the 9600gt dies at that resolution.


If you bothered to actually read my post you will notice that I have provided a link to the review site where those benchmarks came from. They test at 3 other lower resolutions.
a c 358 U Graphics card
March 18, 2009 11:00:24 PM

jennyh said:
I don't consider simply taking this eco 9800gt at 10% less performance proves anything. Anyone can pull numbers out of their behinds.

Why assume a 10% drop in performance? Let's take the previous TDP of 105w and compare it to the current TDP of 75w. That's a 30% reduction in power for only 10% reduction in performance? If Jaguar can do that, he should be working for Nvidia.

What if the number is closer to 20% instead?



If you bothered to actually read my post you would have noticed my statement that the ECO's GPU is 50MHz slower which is a 8.33% decrease and shader speeds has been reduced by 10% compared to a stock 9800GT.

TDP refers to heat generated, not power consumption. Decreasing the die size and lowering voltage will lower TDP.

Stop being a dumbass and f*u*c*k*i*n*g read.
March 18, 2009 11:13:00 PM

LOL

Jaguar FTW.

K close the forum we're don here Moderators:D 
a b U Graphics card
March 19, 2009 12:44:03 AM

jaguarskx said:
If you bothered to actually read my post you will notice that I have provided a link to the review site where those benchmarks came from. They test at 3 other lower resolutions.


How about you actually show us the results at all resolutions instead of at 1920x1200 'because that stresses the cards most'.

While you're at it, prove to us that this eco 9800gt is 10% slower just because you pulled that number out of your ass?
!