gts 250 reviews

[Razor1]

http://www.tomshardware.com/reviews/...-250,2172.html
http://www.guru3d.com/article/palit-...-review-test/1
http://hothardware.com/Articles/NVID...am-GPU/?page=1
http://www.pcper.com/article.php?aid...e=expert&pid=1
http://www.hexus.net/content/item.php?item=17411
http://techreport.com/articles.x/16504
http://www.driverheaven.net/reviews.php?reviewid=725
http://www.anandtech.com/video/showdoc.aspx?i=3523&p=1

clearly matches the 4850 in the games used, and its extra memory comes in very handy.

It also has lower power consumption then the 9800+ while having double the memory so its not the exact same card, a new respin with benefits, maybe but definitly has benefits over the 9800+

 

[entity279]

On these reviews, it's safe to jump directly to the conclusion

Ok, so we have a price reduction that makes it more attractive to buy than the 4850 for now. It's on the plus side (rather than minus) for the "pretty informed" consumer.

 

[compres]

So this is the superclocked version with 1100 mhz memory. Interesting.

 

[iwod]

Well, no, the best buy now is 4870 which is priced at $149.

Although $129 GTS 250 512Mb still beats 4850.

I wonder when will the 40nm GTS250 comes out.

 

[entity279]

Makes you think who can win this price/performance war in the 100$-150$ range.

I personally think AMD has a significant upper hand with the upcoming introduction of 40nm parts. Moreover, since the 4870/4850 was designed to face only g92 parts, with maybe a redesigned, cheaper PCB they could go further down. Ofcourse, these moves take time.

 

[neliz]

So, these are the reviews with the overclocked cards which will not be $129, because that's the price point for the non-overclocked cards coming in two weeks.

Unless someone wants to step up and tell me that nV will sell the non-OC GTS-250 for $99, because they would have a winner there.

It's good to see that only THW and Guru3D didn't sell their soul to nV marketing. THW gives CUDA and PhysX the credits it deserves (none) and G3D review a game that's on the "no-go" list, but they were only allowed to because they have a 2GB model there.

 

[neliz]

I personally think AMD has a significant upper hand with the upcoming introduction of 40nm parts. Moreover, since the 4870/4850 was designed to face only g92 parts, with maybe a redesigned, cheaper PCB they could go further down. Ofcourse, these moves take time.

The "costdown" boards are already on the market in Asia, they haven't hit the US yet, but I bought my 4850 "costdown" in taiwan for less than $100.

 

[Razor1]

So, these are the reviews with the overclocked cards which will not be $129, because that's the price point for the non-overclocked cards coming in two weeks.

Unless someone wants to step up and tell me that nV will sell the non-OC GTS-250 for $99, because they would have a winner there.

It's good to see that only Guru3D didn't sell their soul to nV marketing and review a game that's on the "no-go" list, but they were only allowed to because they have a 2GB model there.

PC Perspective, Tech Report, Driver Heaven, and Guru 3d (guru and had a card with a 2 mhz overclock on the core and 2 gigs of vram), are all stock for the 1 gb version which go for $149, Guru's might be more expensive because of the extra gb of ram.

 

[neliz]

PC Perspective, Tech Report, Driver Heaven, and Guru 3d (guru and had a card with a 2 mhz overclock on the core and 2 gigs of vram), are all stock for the 1 gb version which go for $149, Guru's might be more expensive because of the extra gb of ram.

So how much will the non-oc cards cost then? $139? as of now the 1GB cards on review won't be the cards you buy for $149 when they stock up at retailers. It's a $150 pissing match which the GTS250 can't win against the 4870.

 

[Razor1]

The non overclocked ones with 1gb come in at $149 for now, after the AMD price cuts they will certianly have to drop price, and with the 512mb version comes in at $129 right now.

 

[neliz]

The non overclocked ones with 1gb come in at $149 for now, after the AMD price cuts they will certianly have to drop price, and with the 512mb version comes in at $129 right now.

Sucks to lower the prices on your lower priced products before they even hit the market. Especially since you had to spend money on flashing and re badging them

 

[Jawed]

It also has lower power consumption then the 9800+ while having double the memory so its not the exact same card, a new respin with benefits, maybe but definitly has benefits over the 9800+

1GB 9800GTX+ cards have been around for a while at exactly the clocks of GTS250.

If it's a respin will the chip labelling indicate this?

Jawed

 

[neliz]

1GB 9800GTX+ cards have been around for a while at exactly the clocks of GTS250.

If it's a respin will the chip labelling indicate this?

Jawed

I think it was pretty obvious that there is no re-spin going on: G92-420-B1 or G92-421-B1 depending on if you get a rebadged or a new GTS250..

Lower powerconsumption (a handful of Watts) can be contributed to a new board layout. Heck, the [H]'s 260vs4870 review shows you can save almost 15Watt by just replacing the cooler.

Take a look at the powerfigures from hexus and TH
http://www.tomshardware.com/reviews/geforce-gts-250,2172-10.html
http://www.hexus.net/content/item.php?item=17411&page=11

TH shows identical power usage to the 9800GTX (albeit a 512) and lower PowerConsumption than a HD4850 both under idle/load (~15W on both counts)
Hexus shows a 25W saving over a GTX+ and 14W over a 4850 at idle, yet the 4850 consumes 25W less under load while the GTS has about the same (17W) difference against the GTX+
Hexus tests under 3dM.
G3D saw 2D clocks not working on the 250 and HotHardware has the GTS and 4850 within 10 watts, where the 4850 is actually more efficient at idle.

So, It's marginally better powerconsumption due to board layout.

 

[Razor1]

1GB 9800GTX+ cards have been around for a while at exactly the clocks of GTS250.

If it's a respin will the chip labelling indicate this?

Jawed

It should, I'm just guessing on the respin, the power consumption is really way down, the board redesign could cause it to go down, but I'm suprised its down 20-30%, that is quite high.

 

[Psycho]

Lower powerconsumption (a handful of Watts) can be contributed to a new board layout. Heck, the [H]'s 260vs4870 review shows you can save almost 10Watt by just replacing the cooler.

You can also get lower power consumption by using handpicked cards (maybe even at lower core voltage) for the reviewers...
Regarding the 4870 idle power I think it's a question of new vs old revisions of the card and bios - just take a look around the new reviews, in some of them it's very close to the gtx260 55nm, in others it's much more. Maybe AMD need a rebrand/relaunch

Btw could be interesting to see a board comparision between this and the 8800GTS. The 9800gtx was beefed up to accomodate the higher current etc of the OCed g92a, and thus overkill for the g92b (which is what they have fixed now), so I wonder how close this new revision really is to the original g92 board. Any review with naked board shots?

 

[Jawed]

I think it was pretty obvious that there is no re-spin going on: G92-420-B1 or G92-421-B1 depending on if you get a rebadged or a new GTS250..

421 v 420 is purely binning, isn't it? So that indicates that the "highest performance" GTS250s are simply using cherry-picked chips.

Lower powerconsumption (a handful of Watts) can be contributed to a new board layout. Heck, the [H]'s 260vs4870 review shows you can save almost 15Watt by just replacing the cooler.

Yeah, that's a very good point: a cooler chip will actually consume less power.

One of the issues with a lot of reviews is that they use data from the board they received way back when. A current 9800GTX+ board may well have better performance simply because it's 6 months newer.

Jawed

 

[neliz]

One of the issues with a lot of reviews is that they use data from the board they received way back when. A current 9800GTX+ board may well have better performance simply because it's 6 months newer.

Jawed

Indeed.
http://enthusiast.hardocp.com/article.html?art=MTYyNiw5LCxoZW50aHVzaWFzdA==

What was true six months ago (4800 launch) Radeons are hotter and consume more power than geforces. but now they're plain equal, heck reference Radeons now run cooler and power consumption is basically the same (~15W variance), all features the GT200 won half a year ago. It goes to show that one of the reviews here lists a 4850 as using 25W more under idle&load than a 4870 and they're comparing those numbers to a GTS250 :|

 

[Jawed]

Yeah, that's a very good point: a cooler chip will actually consume less power.

Hmm, the tests at HardOCP don't support this. The article attributes the power difference to the cooler.

Jawed

 

[neliz]

Hmm, the tests at HardOCP don't support this. The article attributes the power difference to the cooler.

Jawed

AMD Radeon HD4870 temp/power figures, both tests used furmark, I don't have a power figure on the new test system:

Launch:
Temp: 64/84 (idle/load)
Power: 187/359 (109W system)

Now
Temp:37/61 (idle/load)
Wall: 178/336

A shame they didn't show power/temp during launch reviews of both 260's.

http://enthusiast.hardocp.com/article.html?art=MTYyNiwyLCxoZW50aHVzaWFzdA==
http://hardocp.com/article.html?art=MTUyNCwzLCxoZW50aHVzaWFzdA==

 

[Jawed]

http://www.bit-tech.net/blog/2009/03/03/where-is-bit-tech-s-geforce-gts-250-review/

While we wait for our card to arrive, we're going to be playing with our GeForce GTS 250 TipeXXX Edition... It's wicked fast.

More seriously:

[...]to our surprise, the general consensus from partners was that they were prohibited from sending samples to bit-tech and Custom PC before launch.

I vote we get this thread renamed: "GTS 250 TipeXXX Reviews". It's like being in school all over again

Jawed