HIS Radeon HD 5770 1 GB

[W1zzard]

he is talking about the pictures behind click links .. they are 800x600 now, and he wants 1024x768 which sounds reasonable. i'll look into it
thumbnails wont be made any bigger for the forseeable future

 

[entropy13]

W1zz, is the 5770 really a bit wider than the 5850/5870 or is it just an optical illusion and the difference in length just makes it appear that it's wider (with the 5770 being shorter than the other cards)?

 

[W1zzard]

W1zz, is the 5770 really a bit wider than the 5850/5870 or is it just an optical illusion and the difference in length just makes it appear that it's wider (with the 5770 being shorter than the other cards)?

5850 and 5770 seem to be 2 or 3 mm wider than 5870, nothing worth mentioning. i wouldnt have noticed it

 

[mdm-adph]

Is that overclocked memory really sitting at 6000MHz effective speed?

The "artifact eliminator" wasn't sacrificing performance at that speed?

 

[PVTCaboose1337]

Yeah this creeped on to newegg last afternoon, woke up, and a review appears! Card appears to be nothing good, maybe just a tad better than my 4850, but really, feels kinda worthless IMO.

 

[entropy13]

5850 and 5770 seem to be 2 or 3 mm wider than 5870, nothing worth mentioning. i wouldnt have noticed it

Well that's weird. I see no difference between the 5850 and the 5870, but the 5770 really looks almost a cm wider than the two, even after a second look. Must be just an illusion due to length differences as well.

 

[jessicafae]

Card appears to be nothing good, maybe just a tad better than my 4850, but really, feels kinda worthless IMO.

I was thinking the same thing, but just could not bring myself to say it. These cards only make sense to someone who skipped the entire 48xx and GT200 generations. Or someone building a whole new computer and wanting mainstream gaming capabilities. The 4770/4850/4870 will soon be out of the channel and these will be the only options. They are good replacement cards, but not exciting in anyway.

 

[W1zzard]

Is that overclocked memory really sitting at 6000MHz effective speed?

The "artifact eliminator" wasn't sacrificing performance at that speed?

yes, it scales to that and is stable, i know what i'm doing

 

[Delta6326]

i really like these cards they are nice and low power but ima stick with my CF 4850's
Did any one else see that the ATI chip on the bottom left pic is upside down? maybe thats why she overclocks so good

 

[DonInKansas]

This card is awesome at the price point. Sips the juice and can play with the big boys. I lurve it.

 

[[I.R.A]_FBi]

dugg

 

[theorw]

WOW @meme OC!!
Would it be 512 bit and u d have 384Gbps bandwidth!

 

[wiak]

i've some problem with the forum..if 'm not wrong W1zzard asked to me something,but now i can't see his post

We talked about bandwidth...
The 5770 is bandwidth limited because has the same specs the 4890 has, except bus and memory clock; the 4890 has a memory bandwidth of 124,8 GB/s, 5770 76,8...too little, in fact as you can see also in this review: http://www.anandtech.com/video/showdoc.aspx?i=3658&p=12 performance are often worse than 4870...

anandtech uses the 1GB 4870 unless am in the woods (hello)

 

[wiak]

many laptops are still 1024x768. Also all the netbooks and UMPC devices have small screens too. Having a hi-res click link is probably better. The nasty part of the internet aiming at the least common denominator.

but it will still fit wont it (as a wallpaper)
most browsers will resize bigger images to smaller when you view a image in a new tab etc

 

[Benetanegia]

anandtech uses the 1GB 4870 unless am in the woods (hello)

These cards have 1GB too, so that's more like apples to apples. In Wizzard's charts the 5770 isn't faster than 4870 512 MB either (except at 2560x1600). Anyway, the specs of the card, sans the memory bandwidth, should make this card as fast as the 4890 and not the 4870.

It's memory bandwidth limited. The fact that the OC benchmark show a greater performance gain (11,4%) than the core OC (9%) also suggest the bottleneck. Look at any other review, the performance gain usually never gets close to the OC.

 

[W1zzard]

if those cards really were memory starved then wouldnt reducing the core clock result in no performance decrease at all?

 

[angelkiller]

if those cards really were memory starved then wouldnt reducing the core clock result in no performance decrease at all?

I think you're right. Reducing the core would have little effect on performance because the core is waiting on data from the memory. (assuming they are in fact memory bottlenecked)

However, has this been tested? (Did I miss something)

Anyway, you don't contradict Benetanegia. Both of what you say can be true. It's just that W1zzard's test would be a conclusive test. (Can we get a quick test?)

 

[W1zzard]

i'm just asking if we agree with this testing method .. i can run some benchmarks later .. right now g 210 is running in the vga benchmark rig

 

[angelkiller]

Do some test at stock speed. Do a core only OC. Then do a mem OC only. And you can toss in a core+mem OC just for comparative purposes. That would show how much of an effect each OC has on performance. So the % gain vs % OC. [EXAMPLE] If you see a 10% gain from a 25% mem OC only and a 10% gain from a 10% core OC only, then you can see that OCing the core is more effective.

The underclocking is basically the same thing in the other direction. Your test of just underclockign the core will show how memory limited the card is.

Hopefully that makes sense.

 

[Paintface]

nice cards and reviews as always , as rule of thumb ignoring the DX11/low heat-power of the new cards and considering worst case scenario for pure performance.

4770 = 4850 = 5750

5770 = 4870

 

[jagd]

I think there is a tricky part for test w1zzard ,shader clocks .youll reduce shader clocks at sametime when you reduce gpu clock or bump shader clock when you OC gpu.How will this effect results?

Edit : a link to 9.10 drivers posted by Thrackan it will be there soon probably (not ready for download yet) http://support.amd.com/us/kbarticles/Pages/ATIRadeonHD5800seriesrecommendedgraphicsdriver.aspx

 

[theorw]

I think there is a tricky part for test w1zzard ,shader clocks .youll reduce shader clocks at sametime when you reduce gpu clock or bump shader clock when you OC gpu.How will this effect results?

Edit : a link to 9.10 drivers posted by Thrackan it will be there soon probably (not ready for download yet) http://support.amd.com/us/kbarticles/Pages/ATIRadeonHD5800seriesrecommendedgraphicsdriver.aspx

what difference would this make?Since shader clocks run at the same speed as the core?

 

[Steevo]

That is some insane memory OC.


Was the GPU at stock voltage for the overclock, and if we might ask what voltage does it show in the drivers on on the card if you have the voltage test points figured out? Do you plan on redoing some of the benchmarks with DX11 titles when they hit?



Awsome review, one of the biggest reasons I was drawn here, and remain.

 

[W1zzard]

as the page says: stock fan and voltage settings

not sure about dx11 yet .. how would i compare nv dx10 w/ ati dx 10.1 w/ ati dx 11 ?

 

[zithe]

I'd like to get a couple. Would be a nice move from my current card.