Sign in with
Sign up | Sign in
Your question

4870/4850 "official" benchies....they're here....

Last response: in Graphics & Displays
Share
June 11, 2008 3:45:58 PM

looks like fud 2 me
June 11, 2008 3:46:47 PM

Average 25-45% faster then 9800GTX... Damn i hope these are real stats.
Related resources
June 11, 2008 3:49:37 PM

SpinachEater said:
.....but from AMD...and with no details....


That one line makes the results questionable at best. IF the card perform as advertised, then they will be a nice step up. But this is the next generation ATI card against today's generation Nvidia card. Until there are benches to compare it to Nvidia's new card, the comparison doesn't mean all that much. And being from AMD, its suspect anyway. Until benches come out from independent testers, its all speculation.
June 11, 2008 3:52:02 PM

At least the y axis starts at 0.

The secrecy surrounding these cards amazes me considering how much we know about Nvidia's new cards.
June 11, 2008 3:52:20 PM

I am thinking real, and AA may finally be 'fixed'. Maybe there are 800 SPs w/ a good portion dedicated to AA.

What I really am wondering is if the 4870 will be on par w/ the GTX 260.
June 11, 2008 3:56:17 PM

Oh sweet jesus.

Please be true.

The last "power" card I had was a Ti4200 (currently x1550) - it's time for a new powerhouse card. :D 


If the HD4850 can beat the 8800GT, and the 9800GTX is not that much faster (~20%) - then looking at that graph, the HD4850 should be on par or quicker than a 9800GTX...
June 11, 2008 4:08:30 PM

I just hope that ATI has fixed their historic driver problems. Thats what originally turned me off from ATI cards after my 9800 pro
a c 84 U Graphics card
a b À AMD
June 11, 2008 4:09:04 PM

homerdog said:
At least the y axis starts at 0.
My thoughts exactly!

I think those are legit enough, at this point anyways...
And if they OC as advertised... :whistle: 
June 11, 2008 4:25:41 PM

where is the ape when you need him.i hope he can shed some thoughts.i really like to read his posts
June 11, 2008 4:28:45 PM

sciggy ati with driver problems....... you obviously never owned a 6600 gt or a 7800 gtx :( 

after those two cards im a firm believer that every new product launch WILL have its problems
a b U Graphics card
June 11, 2008 4:32:47 PM

That, and the fact that nVidia was almost sued for not having DX10 drivers
June 11, 2008 4:34:09 PM

I hope these are legit. I plan to pick up a 4870 at launch.
June 11, 2008 4:38:16 PM

Slightly OT, but what kind of price cuts can we expect on the current gen Radeons (3850, 3870) when these come out.

I'm looking to get into pc gaming btw. Then again if the 4850 is cheap enough who knows.
a b U Graphics card
June 11, 2008 4:52:23 PM

macer1 said:

after those two cards im a firm believer that every new product launch WILL have its problems


Yeah that's been my belief since like the Matrox Millenium era. :pfff: 
Almost every new card needs a bit of time for user feedback to tell the driver team "oh yeah you didn't think we'd play this, or do this with the card, but voila, bug/error".

Anywhoo, hopefully with the architectures being more similar to the previous generation and the maturity of Vista, that it'll be much quicker this time around for both.
June 11, 2008 4:54:27 PM

Heck i still use my ti4400 still get good fps but no dx:( 
June 11, 2008 4:59:28 PM

Those results look reasonable and I think it is fair to compare the HD 4870 with the 9800GTX. That is the card it will be competing against- similar price tag (or so we're told!). ATI never said that the HD 4870 would compete with the GTX 280-hell, it's gonna be half the price! As for the GTX 260, maybe the 4870 won't be far off... (and for a much better price too!)
June 11, 2008 5:03:59 PM

I would have like 4850 vs 9800gtx and 4870 vs 9800gx2.

Still no real numbers. More fud.
a b U Graphics card
June 11, 2008 5:05:07 PM

area61 said:
where is the ape when you need him.i hope he can shed some thoughts.i really like to read his posts


ASCII and I shall appear... :sol: 

My 2 issues with these benchies are;

A) from a presentation point of view I'm a little skeptical due to the positioning of the benchmarks overtop of the page numbers, however anyone can be sloppy in presentation software so they could be legit.

B) Even if legit, these are definitely the best case scenario so need a good dose of NaCl ...... however that the benchies show high resolution, with 4X AA, and with the DX10 version of some benchies to me either shows someone's fantasy of improvement, or if legit that AMD is doing something different with their AA.

Unlike thes first round of benchies with no AA, this would focus on the area that ATi used to be good at, but totally surrendered in their HD2/3K products performance (while maintaining and improving their AA quality that they were also known for).

So IMO promising, but still, can't take them at face value because of the two major issue that I doubt their autheticity, and that they're overly glowing whith not only nary a negative benchmark result in the group , but none that are like 'hey we improved enough this round to get a slight lead', everything is over 25% improvement on immature drivers. Which, even if true, leaves me with the question: So what's the bad news? Everything has a weakeness, I wonder what it is, is it major or insignificant? :heink: 

I wonder what the inverse slide looks like, something that ATi might pass around (internal only) as 'things to look at immediately and see if we can improve with drivers or something', I think I'll be most interested in the GTX2xx and HD4Ks weaknesses at first, because that will give alot of insight into their design too.
June 11, 2008 5:08:52 PM

Looking at those charts again, with AMD listing all the Nvidia cards at "1", it implies that the 4870 is about 42% faster than a 9800GTX in Crysis. That sounds good, but without actual numbers of the FPS of each card, we don't know what the 42% actually represents. Besides that, since this is from AMD, we might keep in mind how good AMD used to claim the K10 CPUs would be, and what really happened when they were released, or how great the 2900 cards would be, and what really happened. No, I don't trust these charts and will I'll wait until real websites, Tom's, Anandtech, etc do real benches and then I can see how well they perform in the wild before I get too excited.
June 11, 2008 5:13:44 PM

why need AA when you can just crank up the resolution settings? >:D 
June 11, 2008 5:20:29 PM

Simply put 4870 will probably do Crysis @ very high long as you keep the resolution down to 1600x1200 or lower is my guess without AA and 4850 will do 1280x1024. 9800gtx comes close but it still doesn't have enough umph to pull very high.

I think 4870 and 4850 will do very well simply the cost and power consumption factor. Something in the lines of G92 and GT200 performance.
June 11, 2008 5:25:12 PM

How much are these cards supposed to cost? With the 9800GTX being a relatively reasonable $300, these cards should be not TOO much more.
June 11, 2008 5:26:45 PM

We get all kinds of numbers but the price is somewhere from low $200 for 4850 and mid $300 for 4870.
Anonymous
a b U Graphics card
June 11, 2008 5:27:10 PM

I doubt these are real... anyway as TGGA said its best case scenario... and there were already nvidia charts like this (most definitely fake) that showed like a 100 % increase... i have a feeling the cards actual performance increase over the 9800 gtx is around 15 - 20 %... maybe that will increase with drivers but for such a massive step up in a generation... they literally doubled their top end offering in one generation... that seems a little steep to me

15 - 20 % would be nice increase in performance... considering i have a 8800 GTS 640 that is oced... I'm thinking i'd get a performance increase of around 30% and maybe more with the X2...

But I just wanna see real benches comparing the 4870 x2 to the gtx 280

cause i'll get the better of the two... unless the performance difference is negligible then i'll get the cheaper of the two... I recently got a 24 inch monitor so i need the horsepower to drive it
June 11, 2008 5:29:44 PM

I'm pretty sure a 9800GTX could drive a 24" without many problems.

Yea I'm thinking the same way...I want to see what the GTX 280 can do. The release date on the 4870X2 is too far away though, I need to build before then.
June 11, 2008 5:48:31 PM

Quote:
I doubt these are real... anyway as TGGA said its best case scenario... and there were already nvidia charts like this (most definitely fake) that showed like a 100 % increase


That was my sarcastic point exactly. It is a biased set of data if it comes from AMD so even if they are 100% legit, they are almost meaningless. In addition, the Y scale is pretty meaningless without any label...1.4 what? FPS? %? Apples? Watts? It is a graph left wide open for the intention of people to draw their own conclusions. It is nice to see some "official" performance leaks though, truthful or not. Who knows though...this could be another photoshop job.
June 11, 2008 6:02:10 PM

aznguy0028 said:
why need AA when you can just crank up the resolution settings? >:D 

thats kinda hard if u have an lcd & r already running at ur monitors native res and hate jaggies like me :)  lol
June 11, 2008 6:03:53 PM

SpinachEater said:
That was my sarcastic point exactly. It is a biased set of data if it comes from AMD so even if they are 100% legit, they are almost meaningless. In addition, the Y scale is pretty meaningless without any label...1.4 what? FPS? %? Apples? Watts? It is a graph left wide open for the intention of people to draw their own conclusions. It is nice to see some "official" performance leaks though, truthful or not. Who knows though...this could be another photoshop job.

1.4 times the performance
a b U Graphics card
June 11, 2008 6:09:36 PM

1.4 times the marketing hype....

1.4 times the Bling...

1.4 times the Satisfaction... MMmmm satisfying !

1.4 times the power consmption / heat production...

1.4 times as much noise...

1.4 times as much eye strain from trying to look for IQ differences between cards...

1.4 times the Bling... oh wait I already said that. :sol: 
June 11, 2008 6:13:55 PM

Hahahaha, exactly

jjokubauskas said:
1.4 times the performance


A very wise prof that I once had always used to say, "keep your eye on the doughnut, not the hole" when it was exam time. Don't fall into the hole my friend.
June 11, 2008 6:14:05 PM

I have never seen a benchmark with all games showing equal margins with the same hardware 1.4 in crysis and 1.3 in doom3, or they should label the X and Y axis...
June 11, 2008 6:20:24 PM

kooldj said:
I have never seen a benchmark with all games showing equal margins with the same hardware 1.4 in crysis and 1.3 in doom3, or they should label the X and Y axis...



Yeah, DX10 Crysis shows higher Y axis (or insert label here) than DX9 Fear? I am thinking...no ,but I am not familiar with how taxing Fear is.
June 11, 2008 6:23:15 PM

It just means that the 4870 beats the 9800gtx more at crysis than it does at fear.
June 11, 2008 6:25:55 PM

That doesn't seem logical to me....but like I said...I am not familiar with Fear

Edit: Do NV cards do really well in Fear?
June 11, 2008 6:27:23 PM

These benchmarks I admit seem little fishy, because they seem suspiciously consistent, but that doesn't mean they are fake.

If they are truth (yey :)  ) then it will seem ATI have fixed a lot of issues for real and now we can expect some sustainable performance not so dependable on drivers probably as 2K/3K series

I am positive about them this way :) 
June 11, 2008 6:30:21 PM

kooldj said:
I have never seen a benchmark with all games showing equal margins with the same hardware 1.4 in crysis and 1.3 in doom3, or they should label the X and Y axis...


Are you stupid or you all play stupid. Enough with the stupid questions I am getting nervous.

1.4 on the scale means ATI gets 1.4x fps in that particular game compared to the 9800GTX

So to say it simple as to a 9 year old if 9800GTX gets 30fps on Crysis then ATI 4870 gets 30x1.4=42fps

I thought simple math is mandatory in school :pfff: 
Anonymous
a b U Graphics card
June 11, 2008 6:35:35 PM

no the games like fear and what not are not taxing... but you can max that game out at 2560 by 1600 on like a 8600 gt... (probably not but w/e) and theres an fps cap... so if the 9800 gtx gets 97 fps... you can only increase it so much... and even then its negligible because the eye will never tell the difference
Anonymous
a b U Graphics card
June 11, 2008 6:36:10 PM

^ i know thats not accurate at all... but it makes my point
June 11, 2008 6:38:36 PM

rawsteel said:
Are you stupid or you all play stupid. Enough with the stupid questions I am getting nervous.

1.4 on the scale means ATI gets 1.4x fps in that particular game compared to the 9800GTX


It is assumed that it is performance but since it is normalized data lacking a Y scale unit, you can't jump to a conclusion. It could be price/performance ratio for all we know. It is up in the air without supporting text from AMD/ATI. I am going with Y = fun factor. It is 1.4x more fun to play it on the 4870. :sol: 
Anonymous
a b U Graphics card
June 11, 2008 7:12:30 PM

i'm going with easiness to pwn noobs in the games vs 9800 gtx

1.4 times easier to pwn noobs with a 4870 then a 9800 gtx =D
June 11, 2008 7:15:09 PM

kooldj said:
I have never seen a benchmark with all games showing equal margins with the same hardware 1.4 in crysis and 1.3 in doom3, or they should label the X and Y axis...

Well it probably isn't 40% faster in ALL games.
a b U Graphics card
June 11, 2008 7:18:27 PM

TGGA, some people are saying that you can tell the shader count by the perlin noise tests in both 3DMark06 and Vantage. Using this, and doing the math, theyve come up with 800 shaders. Now this is all above my head, but is there any credence to it, or could there be also other influences not seen/accounted for?
June 11, 2008 7:29:47 PM

SpinachEater said:
It is assumed that it is performance but since it is normalized data lacking a Y scale unit, you can't jump to a conclusion. It could be price/performance ratio for all we know. It is up in the air without supporting text from AMD/ATI. I am going with Y = fun factor. It is 1.4x more fun to play it on the 4870. :sol: 


It says pretty clear "PERFORMANCE". Are you blind or what. What is the purpose of Price/Performance for each game separately. Don't try to search for a veal under the bull.
June 11, 2008 7:47:45 PM

SpinachEater said:
It is assumed that it is performance but since it is normalized data lacking a Y scale unit, you can't jump to a conclusion. It could be price/performance ratio for all we know. It is up in the air without supporting text from AMD/ATI. I am going with Y = fun factor. It is 1.4x more fun to play it on the 4870. :sol: 


RawSteel, you are making a fool of yourself calling people stupid when this guy has a point. The sly presentation of this graph makes it pretty uninformative and Spinacheater's criticisms are completely valid. A rudimentary understanding of graphs, and their manipulability, would lead you to the same conclusion.
June 11, 2008 7:51:19 PM

rawsteel said:
It says pretty clear "PERFORMANCE". Are you blind or what. What is the purpose of Price/Performance for each game separately. Don't try to search for a veal under the bull.


Performance is a blanket term that can have many interpretations (FPS, efficiency, etc.), making the metric meaningless. They should have specified. The fairly uniform performance gains across games is also a bit suspicious/silly.
June 11, 2008 8:01:03 PM

scooterlibby said:
RawSteel, you are making a fool of yourself calling people stupid when this guy has a point. The sly presentation of this graph makes it pretty uninformative and Spinacheater's criticisms are completely valid. A rudimentary understanding of graphs, and their manipulability, would lead you to the same conclusion.



I am being semi-sarcastic. The biggest factor is that if it comes from AMD/ATI, it is a biased assessment and has to be taken with a pound of salt. The audience for the graph is most likely investors and AMD will give them what they want to see...all of the companies do this including Nvidia and Intel.
June 11, 2008 8:04:28 PM

I am just saying that either these graphs are fud or they are real , there is not much to think about it. I also think the consistent performance accross all games is suspicious.

The thing is if the charts are true then why would you presume its everything else but not clear performance comparison.

"The most obvious thing is often the truth" some wise guy said long time ago
!