5850 just as fast as a 5870?

[Schmide]

I don't have strange hatred towards 5xxx series just laying some hard facts. I just like dissecting hardware and get in to the nutty gritty of things and not some fud going around forums because someone who you think cool said so.

In case of these multi-account trolls it's obvious Schmide is one. if it made him feel better by hiding his original account and telling me I don't know what I'm talking about with personal references all power to him. I can also guess the other guy. So hilarious.

100% WTF. I know you think I'm out to get you, but ask your local MOD to check the IP's and you'll see I never posted in this thread. While you're at it ask them to review the freaky homophobic PM's I've received from you.

I call shenanigans and ask for punishment.

 

[AzN]

I agree with this

Nice post count. Soon Schmide appears.. lol

 

[Schmide]

Nice post count. Soon Schmide appears.. lol

Hopefully the Mods will appear soon and straighten your paranoid ass out.

 

[AzN]

http://www.firingsquad.com/hardware/ati_radeon_hd_5850_performance_preview/page19.asp

Overclocking tests.

5850 beating out 5870 in all tests and that's just average frame rates I wonder what minimum frame would be like. Mostly bandwidth advantage and slight Pixel fill advantage. 5870 with shader and texture advantage.

5850 core 878/1400
Shader: 1260000 MOperations/sec
Pixel Fill Rate: 28096 MPixels/sec
Texture Fill Rate: 63216 MTexels/sec
Bandwidth 179.2 GB/sec

5870 core 850/1200
Shader: 1360000 MOperations/sec
Pixel Fill Rate: 27200 MPixels/sec
Texture Fill Rate: 68000 MTexels/sec
Bandwidth 153.6 GB/sec

So you see no magic drivers no architectural differences and you've got 5850 beating 5870 swiftly with less shader/texture throughput. 4870x2 is roughly 10% faster than 5870. With the 5870 @ 1800mhz memory same as 4870x2 bandwidth it would easily beat 4870x2 across all resolutions.

 

[AzN]

Do you know how easily I can get a new IP?

Proxy, wifi hacks, and so forth.

 

[Schmide]

Do you know how easily I can get a new IP?

Proxy, wifi hacks, and so forth.

I guess. I would add that the prose is no where near like mine. Honestly, I didn't even read past the first page until maybe 30min ago. I was really surprised when this thread hit 5 pages. The paranoia accounts for a fair amount I guess.

 

[AzN]

that's nice. guys with 2 post to their names.

So funny. Using same words like "instantaneous" to describe in post and in reference to Hardocp which you have in the other thread.

I think we're missing each others point. Clarifying.

Either we accept that the 5770 follows the general frame rate of the other two cards (H graphs), and the frame dips are very instantaneous, or we declare the 3 card's frame rates uncompariable and the dips of the 5770 are dramatic and long lasting.

Any graphs I have seen show very much the same fps curves. If minimum was more than just an instantaneous spike I'd agree, but it is not. There are no long flat parts to indicate memory hogging sections of the game that perform well on another card, it is lower almost across the board.

What makes you think the isntantaneous dips

Yes, he only showed average. But who cares, if you are so set against it then run your own tests with the minimum included. There are many sites that list minimum now. A few sites will even show the full fps log (like Hardocp).

 

[Schmide]

that's nice. guys with 2 post to their names.

So funny. Using same words like "instantaneous" to describe in post and in reference to Hardocp which you have in the other thread.

It's the proper terminology. Daedalus685 does use similar prose as me, but the other guy is no where near mine.

Seriously have you ever seen me do anything sneaky? Despite your paranoia, you know I've always been up front with you.

FFS Daedalus685 has a profile at Tom's going back at least a year. I have a profile over there going back a decade. You'll find my profile on most of the major sites always in good standing.

Edit: Ironically Daedalus685 system is very similar to my game system. Except I have a p45, no SSD, no Creative sound, no Blue Ray, and 2x4870s in crossfire.

 

[AzN]

All the other guy kept saying was you don't know what you are talking about. mocking bla bla bla. Saying the same thing Daedalus was saying. lol

Who the hell uses the word instantaneous? It's the first time I've seen in a thread where 2 posters have used it.

daedalus685 at Tom's? good thing you picked up on that. I suppose that's what you go by there and trolling time here.

Of course I've seen you be sneaky. You are the guy who has a grudge against me and it's been going on for what? 6-12 months? i've even told you to stay off my posts and I would do the same for you because we argue too much and get out of hand.

 

[Schmide]

All the other guy kept saying was you don't know what you are talking about. mocking bla bla bla. Saying the same thing Daedalus was saying. lol

Who the hell uses the word instantaneous? It's the first time I've seen in a thread where 2 posters have used it.

daedalus685 at Tom's? good thing you picked up on that. I suppose that's what you go by there and trolling time here.

Of course I've seen you be sneaky. You are the guy who has a grudge against me and it's been going on for what? 6-12 months? i've even told you to stay off my posts and I would do the same for you.

WTF I got in two arguments with you, never called you names, always stayed on topic. When you disappeared for a while after that, I tried to be cool towards you and you called me a homo. Now you're freaking putting me in threads I never posted in. You need help.

 

[AzN]

WTF I got in two arguments with you, never called you names, always stayed on topic. When you disappeared for a while after that, I tried to be cool towards you and you called me a homo. Now you're freaking putting me in threads I never posted in. You need help.

I've told you to stay away from my threads prior to all this and you kept bugging me through PM. I didn't disappear I was busy for couple days and no posts that interested me which you kept PM and bugging me to post something. I told you stop PM. That's when you called me gurly man and I've returned the favor that you gay.

If you want to talk about hardware no problems. Just stop with the multi-account troll because you will get the sham eventually.

You never posted in? You can deny what you want. So funny that you know daedalos that has a year record of posting at Tom's which you've just started participating in. Is there a Schmide at tom's? Someone ban this multi-account troll please.

 

[Schmide]

I've told you to stay away from my threads prior to all this and you kept bugging me through PM. I didn't disappear I was busy for couple days and no posts that interested me which you kept PM and bugging me to post something. I told you stop PM. That's when you called me gurly man and I've returned the favor that you gay.

If you want to talk about hardware no problems. Just stop with the multi-account troll because you will get the sham eventually.

For one, any post is pubic. I'm not going to curb my postings for anyone.

Since everything is out in the open. This is my PM.

Dude come back. I haven't seen you on in a while.

As much as you seem to dislike me, I think we're more similar than different. Sometimes you respect your foes more than those who agree with you.

Do you dare post your replies??? I know my posts and PMs are respectful and within the rules. Are yours?

I told you straight up I'm not going to curb my posting habits, if you don't want replies. Don't post.

Edit for edit

You never posted in? You can deny what you want. So funny that you know daedalos that has a year record of posting at Tom's which you've just started participating in. Is there a Schmide at tom's? Someone ban this multi-account troll please.

Google serious business.

 

[AzN]

It's called harassment. I've PM and I've told you nicely to stay off my posts prior to your harassment. Multi account is obviously not allowed here to make harassing posts.

Anyways I reported your posts. I'm just going to let the mods take care of this.

Stick to the subject matter. Your reported posts are under review. No need to carry on with this here.
Back OT please.

Anandtech Moderator -Keysplayr

 

[Schmide]

Yes, each card is rendering a frame, and storing it into a frame buffer.. The frame buffer is stored in the memory of the cards, and is doubled so that it is on each card. Do you know what a frame buffer does? It holds the date to be drawn to the display, since only one card can send display information it has to have the entire buffer.

AZN You know my knowledge of rendering. Do you ever think I would dumb down a rendering sequence to the above? Crossfire has been known to implement multi card rendering using 3 different methods.

1) a full surface meaning alternating frames are rendered on each card.
2) a cell based rendering. Alternating chunks, think chess board, are rendered on each card.
3) half frame - like nVidia. One card renders the top the other renders the bottom.

Depending on the method the primary card may contain a chain of 2 or 3 surfaces. Method 1 most likely having 3, method 2 and 3 having 2.

The frame buffer is just the current piece of memory that is being sent to the monitor. During full screen rendering with a flipping chain each surface will take turns at being the frame buffer. In windowed mode a back buffer will be copied to the single frame buffer (primary surface) on the primary card. It would be technically wrong to say both cards have a frame buffer.

Sorry Daedalus685 I'm not saying you're wrong, just not as detailed as I would of been with my own terminology.

 

[coreyb]

ok, so then you overclock the 5870 a lot and what are the results? this kind of thing is possible with every generation of video cards. the 260 could overclock to 280 lvls, etc.

 

[BFG10K]

Then again we have 4870x2 with same amount of cores and lower core frequency that is getting much better results in many games than 5870 with more bandwidth.

This has already been explained as the drivers being the most likely culprit, with a non-bandwidth hardware limitation being the second possible scenario.

If driver optimization happen it will also be applied to RV770.

Nope, not if it targets hardware changes on the 5xxx and/or starts using units that weren’t being used properly before.

That driver improvements would show up in both 5850 or 5870.

Nope, not if the 5870’s excess units over the 5850’s aren’t being properly utilized.

Come on now BFG. In all your tests in your own review. 5770 is beating 4850 in all tests except for 2560x1600 in Serious Sam 2 and COD2. I hardly call that needing driver optimization.

So aside from those two games you’re claiming the 5xxx’s drivers are 100% optimal in all others? That’s a pretty bold statement to hinge your entire argument on. Please provide evidence of your claims, or retract them.

Completely difference architecture that behave differently to some games.

The architecture doesn’t matter if you claim the 5770 is starved for bandwidth. That and the same architecture won’t reveal driver limitations, which is the entire point of this discussion and how it pertains to the 5xxx series.

Again showing only avg frame rates to come with that conclusion. I think I've pretty much answered this in my previous post.

You’ve done no such thing. All you’ve done is selectively picked results whenever it suits your agenda. You’ll accept a minimum (which is a single date data point by definition), but you won’t accept [H]’s extensive benchmark plots which put it into a context and prove you wrong.

This isn't anything new and I've been saying in our very first arguments about core needing right combination of bandwidth. It's just that the some of you can't accept when others try to show you things you don't want to agree with which is the reason why you tested your ultra core/sp/mem differences only reinforces what I've been saying about fillrate and memory. I don't need to write articles to show you when the data is already available on the internet. Which pertains to my other thread. http://forums.anandtech.com/showthread.php?t=2031140

You mean the 8800 Ultra benchmarks which you were happy to quote whenever they suited your agenda? I’ll remind you again that their methodology was exactly the same as it was with my 5770 tests. Now you’re turning around and claiming the 5770 tests don’t count because they only use averages.

So which is Azn? Do averages count or not? Or do they only count when you say so?

I can agree with that tis the same reason why RV8x0 is slightly faster than RV770 and some not.

Then stop running around and claiming the drivers work no differently on the 4xxx series compared to the 5xxx series.

The difference? Mostly smoke and mirrors.

I don’t think you quite comprehend what the changes entail. But then I wouldn’t expect anything different from someone that claims CF/SLI doubles memory bandwidth.

From my extensive testings on all the cards I've owned memory makes dramatic difference in minimum fps and not so much on avg. fps.

What 5xxx cards have you tested in your “extensive testing”? Let’s see your plot points for them. Thanks.

I don't have strange hatred towards 5xxx series just laying some hard facts. I just like dissecting hardware and get in to the nutty gritty of things and not some fud going around forums because someone who you think cool said so.

The 5770 is sitting in my drawer right now because it was purchased exclusively for testing and benchmarking. I have no vested interest in promoting anything about it, except the truth. I really don’t care who buys one, but I do care if people continually parrot misinformation about memory bandwidth.

5850 beating out 5870 in all tests and that's just average frame rates I wonder what minimum frame would be like.

179.2 GB/sec from 153.6 GB/sec is a 17% increase in bandwidth. From that you get 2%-3% change in all three games, which is well within the realm of benchmarking noise. In other words the tests are insignificant and don’t prove anything about bandwidth.

This is elementary to someone who understands the concepts of percentages, and also understands that a minimum is a single data point that could be meaningless benchmarking noise.

 

[ugaboga232]

Azn if you do not stop with your harassment of other users, I will report your posts.

Think about it this way, even if there is double bandwidth, there is double shaders. Thus the card has the same ratio of bandwidth to shader and would still be limited.

Going to an example you have been avoiding, The 5970 which has technically "double" the bandwidth does not achieve double the performance. The same performance problem probably effects it. According to you, it should be as fast as its specs say it should be.

You obviously do not know what xfire is, do not use evidence, and cannot interpret the pieces of evidence you have. Any more of this and I will think you are a troll and report.

 

[Daedalus685]

This went to heck over night..

Anyway AZN.. Does this help?
http://www.tomshardware.com/forum/profile-365837.htm

If you look hard enough you can find my seldom updated sticky in the ATI forums as well..


I have tried to have a civil conversation with you.. granted frustration of your thick headed nature has gotten the better of me in a coulpe comments.. I have atempted to logically explain what is going on to yuo and you refuse to pay it any mind, which is your right.. But don't go slandering me, and by extension someone I don't even know, with paranoid nonsense. If you want to discuss the topic then go ahead, I'll take part. But I will have no further play in this "you are all the same person" garbage.

By the way, what word should I have used instead of:
http://en.wiktionary.org/wiki/instantaneous

I'm a physicist.. it may not be a word you use daily, but it sure is one do (at least lately since am workign on some issues with signal noise).. sorry you find it so offensive for it to be part of my diction.. You might check the thesaurus and find that it is the only technical term that conveys my meaning.. Though I can use "The minimum momentarily drops" but that does not convey the same connotation in regards to the extremely short time I am talking about.

Ssutained minimums matter (Why I would like to see stdev in a lot of reviews), instantaneous minimums mean nothing. You are best to understand the difference.

 

[Meaker10]

I will correct the mistake of saying you only get a max of 50% bandwidth increase at max. It would depend on the balance of reads and writes.

I am not sure but I suspect the cards will do more reading than they will writes.

 

[Daedalus685]

I will correct the mistake of saying you only get a max of 50% bandwidth increase at max. It would depend on the balance of reads and writes.

I am not sure but I suspect the cards will do more reading than they will writes.

It is still only theoretical. I have no idea what kind of optimizations they have in crossfire/sli.. But in the best case the final bandwidth will be (Single card bandwidth * (1 + reads/total memory IO)). Just like with any other system I'd have to guess teh actual cse is far less than that. Though I am not usre what the proportion of read/writes are. I'd assume the same texture will be read multiple times, but other than initial texture writes it is probably 50/50. So it would depend on the game, as to how many times it has to swap textures around. Thus the more memory you have, the less writting you have to do, and the vastly different this math wuold look.

Do you think that it is set up in such a way as to allow near perfect read scaling? I wonder how close it is.. Raid 1 certainly isnt very close to the theoretical read performance increase in any task I've ever played around with. Though the situations are vastly different.

 

[myocardia]

This entire thread shows just what will happen to any forum that is allowed to go for very long without moderation. I think the OP needs a nice long vacation for calling out a member, and on top of that, for "reporting" him, once the called out member finally responds. This won't happen, of course, and who would expect it in a forum with absolutely no moderation??

Please drop the comments about moderation (which don't belong here in VC & GFX), and get back OT.

TIA

Fern
Super Moderator

edit: Sorry GodisanAtheist, I had forgotten that you were the OP. You were obviously not the poster that I mentioned above.

 

[Elfear]

ok, so then you overclock the 5870 a lot and what are the results? this kind of thing is possible with every generation of video cards. the 260 could overclock to 280 lvls, etc.

That's normally how it goes but in this case, the 5850@5870 clocks is only 2-3% slower with current drivers. From the overclocking evidence I've seen, the 5850's tend to overclock a bit higher than the 5870's. With that being the case, the 5850 is basically on par with the 5870, at least until drivers increase the performance disparity between the two (if that ever happens).

 

[hans030390]

I don't have strange hatred towards 5xxx series just laying some hard facts. I just like dissecting hardware and get in to the nutty gritty of things and not some fud going around forums because someone who you think cool said so.[/b]

I've read this whole thread. You're the only one here rambling on about things after obviously being proven wrong by evidence provided by others (you've barely provided any). If these are "hard facts" to you, that is a big problem.

 

[GodisanAtheist]

Holy sweet mother of all that's holy, WHAT HAPPENED TO MY THREAD?!

Alright, can we please let up on Azn and get back on topic (In case anyone forgot, was about the 5850 being nearly as fast as the 5870 @ the same clocks and why this is the case).

I remember someone else on these boards (not in this thread) saying something to the effect of the Radeon's single x32 thread scheduler might be getting swamped or overloaded and ATI would have been much better off going dual x16's or quad x8's or something to that effect. Any thoughts on this?

 

[AzN]

This has already been explained as the drivers being the most likely culprit, with a non-bandwidth hardware limitation being the second possible scenario.

With no proof that drivers are the culprit I might add. Non bandwidth hardware limitation like? When only thing 4870x2 has over 5870 is bandwidth it's the most likely limitation of the 5870.

This is a perfect example of bandwidth not saturating 5870 theoretic peaks while 4870x2 has more bandwidth that peaks more theoretical with less theoretical peaks.

Nope, not if it targets hardware changes on the 5xxx and/or starts using units that weren’t being used properly before.

True that however the hardware changes were mostly artificial between cache and so on that already takes advantage.

Nope, not if the 5870’s excess units over the 5850’s aren’t being properly utilized.

Which goes back to the bandwidth limitation. in the above picture.

So aside from those two games you’re claiming the 5xxx’s drivers are 100% optimal in all others? That’s a pretty bold statement to hinge your entire argument on. Please provide evidence of your claims, or retract them.

These are what 2 7 year old games? It probably never needed any real optimization as it was already fast enough for ATI. Aside from changing a few numbers on their drivers from RV770 to Cypress there really isn't anything different about RV770 and cypress architecture besides tessellation improvements and fetch and so forth.

The architecture doesn’t matter if you claim the 5770 is starved for bandwidth. That and the same architecture won’t reveal driver limitations, which is the entire point of this discussion and how it pertains to the 5xxx series.

Of course architecture matters in this case when the card behaves entirely different to games. You claim 5770 beating GTX285 in 1 game. We have these kind of anomalies going from entirely different architecture to the next. This is nothing new.

You’ve done no such thing. All you’ve done is selectively picked results whenever it suits your agenda. You’ll accept a minimum (which is a single date data point by definition), but you won’t accept [H]’s extensive benchmark plots which put it into a context and prove you wrong.

Again. I've explained to Schmide that I don't trust Hardocp's benches. Reason being... They've had their minimum frames all over the place in the past. It seems they use fraps to record their frame rates but don't get rid of hard drive seeks or margin of errors. I don't see the consistency in their benches.

You mean the 8800 Ultra benchmarks which you were happy to quote whenever they suited your agenda? I’ll remind you again that their methodology was exactly the same as it was with my 5770 tests. Now you’re turning around and claiming the 5770 tests don’t count because they only use averages.

So which is Azn? Do averages count or not? Or do they only count when you say so?

8800ultra had more bandwidth than it really needed and testing minimum fps would be a waste of time. In case of 5770 let me remind you it has much more processing power, 40% texture fill, and nearly same pixel with yet it has 25% lower bandwidth than the ultra.

Then stop running around and claiming the drivers work no differently on the 4xxx series compared to the 5xxx series.

Changing few numbers around 4xxx compared to 5xxx is somehow working differently?

I don’t think you quite comprehend what the changes entail. But then I wouldn’t expect anything different from someone that claims CF/SLI doubles memory bandwidth.

Again mostly prefectching difference between cypress and rv770. BFG tell me what you think about CF/SLI how bandwidth is applied or do you just agree with everyone because someone said so and I didn't? You must have an idea if you don't thin the bandwidth doesn't double.

What 5xxx cards have you tested in your “extensive testing”? Let’s see your plot points for them. Thanks.

Vid architecture act similar to bandwidth and fillrate. you don't need a 5xxx card to get a grasp to get the picture. It's the same reason I was able to tell you that your ULTRA was fillrate limited without even owning one.

The 5770 is sitting in my drawer right now because it was purchased exclusively for testing and benchmarking. I have no vested interest in promoting anything about it, except the truth. I really don’t care who buys one, but I do care if people continually parrot misinformation about memory bandwidth.

I could care less what anyone buys but when reviewers don't test minimum fps but only test avg frame rates to get conclusion and continue to misinform about bandwidth I have a problem with.

179.2 GB/sec from 153.6 GB/sec is a 17% increase in bandwidth. From that you get 2%-3% change in all three games, which is well within the realm of benchmarking noise. In other words the tests are insignificant and don’t prove anything about bandwidth.

This is elementary to someone who understands the concepts of percentages, and also understands that a minimum is a single data point that could be meaningless benchmarking noise.

Again that's just average frame rates with lower processing power and texture fillrate. Now imagine minimum frame rates where that % is multiplied by the bandwidth.

I love how you are quick to discredit me about anything long as you have the urge to agree with the mass. This won't be the first time and I hope it won't be the last.