Sign in with
Sign up | Sign in
Your question

Thuban 1055t vs 965 BE

Last response: in CPUs
Share
August 19, 2010 10:54:33 PM

Lets say you're planning on playing a game the doesn't utilize 4 cores or maybe only uses 2 cores. Will the Thuban outperform the 965? Is the Thubans architecture superior in any way. I was told by someone that they have run both CPUs with a 5870 and saw a 25% gain with the Thuban. Could someone please explain?

More about : thuban 1055t 965

a c 203 à CPUs
August 19, 2010 11:01:16 PM

How do you explain wishful thinking? What game? What resolution? Most games are GPU limited - so that must have been a very interesting game, or maybe using a low resolution or something like that.
Gaming benchmarks X4 965 vs X6 1090T. Check out the WoW benchmark. Then check DAO - it's known to use four cores.
You can plug in the 1055T if you want as well.
m
0
l
a c 203 à CPUs
August 19, 2010 11:06:23 PM

In this X6 1090T / X6 1055T review you can look at gaming benchmarks from a RPG (Anno) FSP (COD:MW2) and SIM (CMR:D irt 2)
m
0
l
Related resources
a c 131 à CPUs
August 20, 2010 12:39:13 AM

Thuban has very few architectural changes from the deneb (965) cores, relatively speaking.
To answer your question, there are no IPC (instructions per cycle) improvements in the deneb core, therefore a thuban processor using 4 or less CPU cores would perform the same as a deneb 4-core processor at the same frequency.
Thuban does have turbo-core going for it, but from what I have read, it doesn't seem to make a major difference.

both CPUs with a 5870 and saw a 25% gain with the Thuban
What game? What settings?
m
0
l
August 20, 2010 10:56:25 AM

I was the person that did the testing. I overclocked both the AMD 965BE and the 1090T to the same 3.91Ghz and ran the same benchmark in iRacing. iRacing is very much a CPU limited game.

5040x1080 triple screen resolution = 25% performance difference in FPS.

The initial Tom's Hardware review that everyone is basing their opinions of this chip was completely flawed. It did not take into account overclocking the CPU. This particular CPU was only at 1.356 V at 3.91 Ghz. It was easy to obtain that speed as well. Since I built it for a client I didn't push it but well beyond 4.0 would have been easy.

This is a much better review of the chip. We can all agree no matter how much you overclock the 965BE it cannot match the results here:

http://www.overclock3d.net/reviews/cpu_mainboard/amd_10...

Be sure to check the gaming section of that review. Hands down it is the best value for the money.
m
0
l
August 20, 2010 11:07:02 AM

I'd also like to add that there were some architecture changes for the 6 core chips. They didn't just slap two more cores on the existing phenom II die. They added a larger L3 cache which seems to help based on my testing.
m
0
l
August 20, 2010 1:05:17 PM

You have spoken. Clearly I have no idea what I'm talking about. :sarcastic: 

Have you actually done these tests yourself? Or did you even read the link that I posted? Either way I'm not on here to disagree with "tech experts" over results that I clearly witnessed myself and documented. I may not have 1,000's of posts here but I'm qualified to run tests and report on their results.

Once overclocked to the same levels on the same system in the same game (iRacing) there was a 25% performance increase. iRacing offloads all of the shadow computations to the CPU and that can really bring nearly any machine to it's knees. It's not efficient.

My findings are consistant with the overclock3d.net review.

I posted here knowing that people that have zero experience with this would be tearing me apart but I'm willing do it. Read the link once overclocked the 6 core is a great chip regardless of the game or how many cores it takes into account and clock for clock it is faster than the 4 core chip.

I comment on this based on actual testing experience not ego or assumed knowledge as I'm reading from you. It is fact, in iRacing I had a 25% performance gain between the two chips.
m
0
l
a c 203 à CPUs
August 20, 2010 1:33:49 PM

the shoe said:
I was told by someone that they have run both CPUs with a 5870 and saw a 25% gain with the Thuban. Could someone please explain?
Running iRacing @ 5040x1080 will get you a 25% gain.
m
0
l
a c 203 à CPUs
August 20, 2010 2:51:27 PM

Tim Nolder said:
My findings are consistant with the overclock3d.net review.
How do you figure that?.
There is no comparison with the X4 955. So overclock3d.net chart gives no clue as to how a X4 955 would improve with overclocking.
X6 1090T results show an improvement OC'd vs Stock.
Dirt 2: 15% (79-90)
NFS Shift: 20% (78-94)
Crysis Warhead: 4% (51-53) If the 1090T OC improved performance 4% the X4 955 OC would improve... 3.2%?
m
0
l
August 20, 2010 3:30:12 PM

I agree there is no data from that review.

Having said that there are plenty of reviews that show the 965BE is no where near the core i7 regardless of clock speed. This shows that the 1090T is on par or better in these applications.

I'm using the core i7 as the constant for comparison.

My reasoning is based on data avaiable core i7 vs the 965BE and core i7 vs the 1090T.

From that I conclude that there is a difference in other applications. If not then I guess the 965BE is equal to the core i7 from the overclocked3d test I linked.
m
0
l
a c 131 à CPUs
August 20, 2010 4:07:48 PM

Ok, so why should I trust Tim's benchmark results over everyone else?

I don't know about the other games, but I do know that Crysis cannot use more than 2 CPU cores, unless there is an update I am unaware of.

Thuban will obviously perform better if the game can use more than 4 CPU cores, otherwise, it is the same, clock per clock, with deneb.
m
0
l
a c 203 à CPUs
August 20, 2010 4:31:45 PM

Tim Nolder said:
My reasoning is based on data avaiable core i7 vs the 965BE and core i7 vs the 1090T.
From that I conclude that there is a difference in other applications.
Do results like these support your conclusion. Or does it tell you anything else about the X6 1055T/1090T vs X4 965?



m
0
l
August 20, 2010 6:34:35 PM

My statement is this. With iRacing at 5040x1050 4x 16x and full shadows I had a 25% performance increase over a 965BE running the same test. None of your data would have predicted that but here we are. I've posted my result in my specific game's forum.

I sell computer systems specifically for that game and I have no interest in misleading anyone. I even ran the tests multiple times.

That is fact. Everything else is my opinion or subjective.

I also posted a result where the core i7 and 1090T are nearly idential when highly overclocked. You can compare it.

Are we to believe that if I overclock a 965BE and core i7 to the same levels and then benchmark a game that relies heavily on CPU usage that they will perform equally? If so then why does everyone consider the core i7 so superior? It is based on synthetics or real world applications?

Lets talk about what is being discussed here. We have tests that prove that the core i7 is fairly equal to the 1090T. You are saying that the 965BE=1090T=core i7. That's the relationship structure.

But we know that the 965BE is not equal to the core i7... so what does that mean?
m
0
l
a c 203 à CPUs
August 20, 2010 9:36:52 PM

It took you quite a while to qualify your 25% performance advantage claim.

Clearly the shoe had some questions about your claim - enough to drop by here and post his question.
It's good to know you weren't trying to mislead him.

"Tom's Hardware review that everyone is basing their opinions of this chip was completely flawed". You should know that overclocking any CPU is always an option - never a requirement. So to say a review is 'flawed' because it did not include overclocking is itself, a flawed statement.

Getting to that 'We have tests that prove that the core i7 is fairly equal to the 1090T' statement?
Tim Nolder said:
Everything is my opinion or subjective.

m
0
l
a b à CPUs
August 20, 2010 10:20:34 PM

Everybody, ignore him, he is in a confused state and thinks the 955 and 1090T are better than the i5 750 in gaming, after being shown 10 benchmarks from 3 sites saying otherwise, and claiming a fanatic AMD fangirl as his source.
m
0
l
August 20, 2010 10:46:35 PM

You're a bad loser and everybody can see it ares1214. :) 
m
0
l
a b à CPUs
August 20, 2010 10:54:23 PM

Ah i see, im a bad loser. Well i am quite bad at losing, so i suppose you are right in some respect. Lets get this straight. You believe the 1090T is better for gaming. I present 3-4 different games, from 3 different sites, and you say they are all biased, I cherry picked, and the 1090T is better. You show 1 benchmark where the 1090T wins (by 1 fps, everything is within about 3 fps) which obviously means it is a GPU limited game, MW2, which makes sense as it is a console game. And im the loser here.
m
0
l
a b à CPUs
August 20, 2010 10:57:30 PM

Oh, lets not forget, the person who hasnt had an intel cpu for over 5 years, me, is all the sudden an intel fanboy for presenting benchmarks showing the 750 winning, and all of the sudden the fanatic paranoid AMD fangirl, jenny, who cherry picks everything to make AMD CPUs look faster, is now the most credible source there is, more credible than THG. And yet im still the loser.
m
0
l
August 20, 2010 11:11:18 PM

ares1214 the only problem here is you aren't understanding the mechanics behind stats.

Yes you showed a few benchmarks of the 750 winning, but Jenny showed what the results were over nearly a hundred benchmarks. If you look at her excel file you can see that there were wins for both AMD and intel chips, but the overall totals cant be denied.

The i7 980 won ahead of all the rest so I think that puts paid to any conspiracy on her part. I agree she was a real fangirl of AMD but she was honest too, if she wasn't I bet she could have manipulated the results to "prove" the 1090T was faster than the i7 980 but she didn't.
m
0
l
a b à CPUs
August 20, 2010 11:21:59 PM

You do realize she even listed the 1055T beating the i7 930 in gaming? Are you really daft enough to believe that?
m
0
l
a b à CPUs
August 20, 2010 11:27:03 PM

You also do realize i could make something EXACTLY like that showing how intel wins? Not too hard. Where as using individual games on reliable sites, such as THG, Anandtech, Cannucks, Bit Tech, Guru 3D and Benchmark Reviews actually cant be cheated.
m
0
l
August 20, 2010 11:35:15 PM

ares1214 said:
You also do realize i could make something EXACTLY like that showing how intel wins? Not too hard. Where as using individual games on reliable sites, such as THG, Anandtech, Cannucks, Bit Tech, Guru 3D and Benchmark Reviews actually cant be cheated.


So do it and show us all.
m
0
l
August 21, 2010 12:09:48 AM

WR2 said:
It took you quite a while to qualify your 25% performance advantage claim.

Clearly the shoe had some questions about your claim - enough to drop by here and post his question.
It's good to know you weren't trying to mislead him.

"Tom's Hardware review that everyone is basing their opinions of this chip was completely flawed". You should know that overclocking any CPU is always an option - never a requirement. So to say a review is 'flawed' because it did not include overclocking is itself, a flawed statement.

Getting to that 'We have tests that prove that the core i7 is fairly equal to the 1090T' statement?


I'm home now and I can actually formulate a decent sentence. I specifically stated what I tested and how I came to my conclusion in the initial post.

When I say it's flawed, it is an enthusiast chip with an unlocked multiplier. It is made to be overclocked and if we are going to talk performance then it needs to be done to see how it compares to the core i7 at the same levels- especially when other sites were making claims at the time that they could get it to 4.0Ghz or greater on a mid level moderate air cooler.

Tom's is one of the most popular sites and I would expect them to run the tests enthusiasts want to see when buying a $300 CPU- someone dropped the ball. Based on reviews from other sites the value of this chip was going to be high overclock potential and it hasn't disappointed. You can take a 1090T and run Asus Turbo V with it and every one that I've done was able to get itself to at least 3.8 or higher. Manual tweaks have easily taken them over 4.0 - something the Phenom II x4's aren't really known for.

I do believe based on my own experience with them that AMD turbo unlocked technology is what holds these chips back at the stock settings. It pegs the hell out of the Vcore raising it to as high as 1.475 to hit 3.8 when at stock speeds. My chip I used in that article needed only 1.356V to hit 3.91Ghz. Additionally with the initial bios versions released for these chips disabling this feature did not actually turn it off. So if you overclocked the CPU would try to push it self even higher by zapping the Vcore. This can be verified in the several hundred page thread at overclock.net on this chip. I did a ton of research on it before I started to overclock it. My research was not all accurate it would seem as someone posted that the L3 cache is bigger and I went with it. My apologies there.

Now that the bios have been fixed (the issue was present on both MSI and ASUS boards I was testing) we are seeing more reviews with data that is contrary to the original ones and showing the upper end of the potential of this chip. That was my point and it was my fault for assuming anyone reading and commenting on this was aware of any of this. My initial testing post on the iRacing forums gave all the details of the testing method and was setup to be replicated for verification with other users. Again I showed a 25% performance increase with those settings in that game. I can't think that's an anomaly. Something about the 6 core chip is causing that.

I knew posting here I was going to be attacked and it was my fault for posting before I could accurately type out a response.
m
0
l
a b à CPUs
August 21, 2010 1:28:29 PM

Who gets 5040x1050 anyway? If you spend that much money on an eyefinity setup, then why wouldnt you get 5760x1080?
m
0
l
August 21, 2010 1:54:15 PM

ares1214 said:
Who gets 5040x1050 anyway? If you spend that much money on an eyefinity setup, then why wouldnt you get 5760x1080?


People who already had 1 or 2 1680x1050 screens?
m
0
l
!