Radeon HD 4770 Released, Industry's First 40 nm GPU

[btarunr]

AMD today released the ATI Radeon HD 4770 graphics card. The release marks several milestones for the company, mainly winning the race for the first GPU to be built on the 40 nm process, and the introduction second-generation GDDR5 memory for the mainstream consumer segment.

The brains of this card is the 40 nm AMD RV740 GPU. Its specifications include 640 stream processors that churn out over 900 GFLOPs of shader compute power, 32 texture memory units, and 16 render back-ends. The GPU is aided by 512 MB of fast GDDR5 memory across a 128-bit wide interface. This provides the same amount of bandwidth as 256-bit GDDR3 commonly found in most graphics card in the range. The card is DirectX 10.1 compliant, and supports the ATI CrossFireX multi-GPU standard.

The card has been launched worldwide, with its initial US price set at $109, and an optional rebate that can send its price further down. In its range, it competes with NVIDIA GeForce 9800 GT, and AMD's own Radeon HD 4830. TechPowerUp is one of the first technology portals to publish a thorough review of the Radeon HD 4770. Our review can be read here.



View at TechPowerUp Main Site

 

[Mussels]

yay for 40nm!

Good price too.

 

[HeadlessChicken]

nice, if i didnt already have a 9800gtx this would look very tempting.

 

[R_1]

Good for ATI, good for us too.

 

[HTC]

Strange: before i saw W1zzard's review, i saw this one: take a look @ the power consumption in it.

Then i read this one and i'm surprised by the huge difference in this aspect of the reviews. I mean: ~10 watts is one thing but there's a difference of 50 in idle and 49 under load.

Dunno if there's a difference between the card AMD sent for W1zzard to test and the card of the other review.

 

[air_ii]

Yeah, it is strange. If I remember correctly, I saw the same thing with reviews of HD4890.

 

[W1zzard]

Strange: before i saw W1zzard's review, i saw this one: take a look @ the power consumption in it.

Then i read this one and i'm surprised by the huge difference in this aspect of the reviews. I mean: ~10 watts is one thing but there's a difference of 50 in idle and 49 under load.

Dunno if there's a difference between the card AMD sent for W1zzard to test and the card of the other review.

different test system, different psu, different testing method. you realize those numbers are for THE WHOLE PC ?

we'll soon have card only power measurements, waiting on a few last parts

 

[HTC]

different test system, different testing method

we'll soon have card only power measurements, waiting on a few last parts

I see. @ first, i thought they were just measuring the card's output wattage but they were measuring the whole system.

The difference is just too big not to notice and it struck me as very odd: thanks for clearing that up.

 

[air_ii]

But different test systems may show large differences in total power consumptions, however, there should be a fairly stable delta between say 4770 and 4850. There can't be that much difference wrt CPU power consumption at the same situation across different GFX cards (although I'm sure there could be some).

EDIT: the numbers that HTC linked to are also for the whole system.

 

[W1zzard]

But different test systems may show large differences in total power consumptions, however, there should be a fairly stable delta between say 4770 and 4850. There can't be that much difference wrt CPU power consumption at the same situation across different GFX cards (although I'm sure there could be some).

exactly

EDIT: the numbers that HTC linked to are also for the whole system.

but our machine is different from what expreview has

 

[air_ii]

Don't get me wrong, I'm not trying to imply anything, I'm just trying to understand. So looking at the results, I'm guessing you're measuring the draw at the socket whereas they could be measuring at the PSU rails (given your systems are fairly similar). Then there could also be the issue of the PSU efficiency at low loads...

All in all, numbers that other sites are giving seem more like it. Especially if you take into account that the 4770 needs only one six pin power cable, while the 4850 needs two.

And thanks for the review, btw !

 

[Gzero]

Can't compare those power figures, TPU uses a completely different system, the overclock will also be different on the cpu. The PSU will have a different efficiency and the hdd is different.

I think W1zzard's review has made it clear (at least to me) that the 4770 just doesn't cut it compared to the 4830.

 

[HTC]

Can't compare those power figures, TPU uses a completely different system, the overclock will also be different on the cpu. The PSU will have a different efficiency and the hdd is different.

I think W1zzard's review has made it clear (at least to me) that the 4770 just doesn't cut it compared to the 4830.

Also, the drivers don't seem to be the same.

Given the reviews thus far presented, i think it's early to say that.

Besides:

we'll soon have card only power measurements, waiting on a few last parts

I'll wait for this.

EDIT

Found a page that links to a bunch of reviews.

There are discrepancies with the power consumption in those as well: check AnandTech and HardwareCanucks, for example.

 

[air_ii]

Can't compare those power figures, TPU uses a completely different system, the overclock will also be different on the cpu. The PSU will have a different efficiency and the hdd is different.

The only part I agree with is the PSU. HDD's power is negligible, really. CPU's overclock is the same in the two reviews (not sure about voltage, etc.), but fair enough, you can't compare it like that directly.

I think W1zzard's review has made it clear (at least to me) that the 4770 just doesn't cut it compared to the 4830.

This is the part I'm not convinced about. Although other sites (Anandtech, Hexus, to name a few) show approximately a 10W higher idle power draw, they present load power draw at some 20-30W less than on 4830...

 

[smartali89]

for sure buy

 

[W1zzard]

This is the part I'm not convinced about. Although other sites (Anandtech, Hexus, to name a few) show approximately a 10W higher idle power draw, they present load power draw at some 20-30W less than on 4830...

our review shows something similar. higher in idle, lower under load. dont worry about 10 Watts here and there.

to the one who said they could measure "at the psu rails" - if someone had the capability to measure like that, they would just measure the card alone. every site who measures "whole system" power measures at the wall

 

[ShadowFold]

On newegg all the cards are using the cheap looking cooler.. I don't doubt it, but it looks hideous.

 

[W1zzard]

On newegg all the cards are using the cheap looking cooler.. I don't doubt it, but it looks hideous.

check my review for what amd had to say, page 3 i think

 

[happita]

I think now that it is on a smaller fab, 40 nm, it will surely overclock up to 4830/4850 potential speeds? I mean, look at the stock core clock of that on the 4850 vs. 4770, 625MHz vs. 750MHz ? That alone is an indication of how fast the GDDR5 is and with the help of controlling heat a bit better through 40nm process. (Basing all this info on TPU's review)
I'm sure that you can OC the core clock even further to about 825-900 and memory clock to about 900-950, then you have a product that is in strong competition with other older 55nm cards. This is in direct competition with not only the 4830 like it says in the first post, but also with the 4850, which is supposed to be the best bargain for the buck. My 2

 

[mdm-adph]

All in all, numbers that other sites are giving seem more like it. Especially if you take into account that the 4770 needs only one six pin power cable, while the 4850 needs two.

4850 just needs one.

I personally love just how small the stock 4770 seems to be, great for small cases.

 

[mamisano]

Over Here it shows power draw of the card, not the whole system.

4770 is listed 15w less consumption at idle compared to 4830, 23w less than 4850
4770 is listed 35w less consumption at 3D load compared to 4830, 60w less than 4850

 

[ktr]

I wonder how well this card scales in crossfire.

 

[btarunr]

I wonder how well this card scales in crossfire.

Scales a nice 82% (all resolutions' average) from Expreview's findings. That's 20% faster than HD 4890, according to the same review.

 

[ktr]

Scales a nice 82% (all resolutions' average) from Expreview's findings. That's 20% faster than HD 4890, according to the same review.

That is shockingly impressive. You can get two 4770 cheaper than a 4890 and be 20% faster...

I may just found my next pair of graphic cards (if I get around building a new rig).

 

[eidairaman1]

just remember dude these are the precursor of what is to be released either later this year or sometime next year, but yeah these are pretty decent.