Why I bought a Bulldozer inside

[jonnyrb]

Somebody on these forums pointed out that Toms and a few other sites used the same benchmarking kit with an nVidia 580 gpu and the same mobo ASUS Crosshair V. This these tests are using a different benchmarking kit and an AMD gpu... *cough* as intended

Nobody can argue with pictures

 

[Chad Boga]

Just upgraded yesterday from my 1100t to an 8120. To me, it was money well spent. Here's why:

Everybody keeps making this bogus claim that today's software/OSes aren't geared as much for multithreading. Those people probably have never looked at their task manager. I have a Logitech G15 keyboard, so I can see per-core usage, at a glance. Windows (and Linux, of course) does an _outstanding_ job of spreading its workload across multiple cores. It means I have a lot more headroom for multi-tasking/threading both now and in the future.

As for games not using more than 2-4 cores? Again: you clearly haven't monitored your per-core usage when playing modern games. I cannot think of one single game that I own which doesn't get all 8 of my cores rumbling. Maybe not all with heavy loads, but working, nonetheless. It has unleashed my crossfire performance...I was very disappointed before.

Now, with all of that said, Sandy Bridge is easily a superior product, overall. I would not recommend BD, in good conscience to most system builders for the foreseeable future - only on a few niches that BD fills well. In that case, overclock an 8120 to get bang for buck. For anybody wanting a solid, future-looking upgrade for their old Phenom IIs w/o changing platforms, the OCed 8120 is definitely the way to go. It will make your computer run smoother in day-to-day workloads, as well as gaming. Just overclock it to or past a stock 8150 - mine is at 4.5ghz @ 1.47v stable. Unless you're doing something like Photoshop that doesn't scale well across cores. Definitely look at the benchmarks to make sure it will fit your needs, but for the gamers, I see no area where it slouches.

Note: these are the only benches I can easily link to right now...Not trying to cherry pick, or even prove anything...Just showing.

8120 @ 4.5ghz:
http://3dmark.com/3dm11/2008319
http://3dmark.com/3dmv/3562222

8120 @ 4.2ghz:
http://3dmark.com/3dm11/2002726
http://3dmark.com/3dmv/3559643

1100t @ 4.125ghz:
http://3dmark.com/3dm11/1902139?show_ads=true&page=%2F3dm11%2F1902139%3Fkey%3DnLK8K6YWbcudhRP3yQS5QK2qHhW38U (This was my system - I just didn't have a futuremark account when I did the bench...Also can't find my vantage score, but it was around 19k for CPU)

I'm surprised you bothered to upgrade from a X6 to a BD 8120, what are you really gaining here other than a lighter wallet?

 

[decode33]

I get more miles per gallon on my v4 japanese car than my v8 american car. I save money.

 

[noob2222]

I get more miles per gallon on my v4 japanese car than my v8 american car. I save money.

you must be one of those that sit in the fast lane trying to pass the european car for 3 miles holding up traffic. "i can't drive any faster, this thing is floored"

 

[alikum]

[url]http://uploadpic.org/storage/2011/ASsfHRT4asGyGFqAWsDB3DfL4.png[/URL]

Very funny. If that happens, enjoy selling kidneys for your Intel parts

 

[wh3resmycar]

just remember that amd fanboyism is built on "AMD will get better in the future (and intel will get worse)".

Just upgraded yesterday from my 1100t to an 8120. To me, it was money well spent. Here's why:

that's not an upgrade.

 

[JAYDEEJOHN]

You guys should read this, it explains why Bulldozer is a bit of a flop:

"Ex-AMD Engineer Explains Bulldozer Fiasco: Lack of Fine Tuning."
http://www.xbitlabs.com/news/cpu/display/20111013232215_Ex_AMD_Engineer_Explains_Bulldozer_Fiasco.html

And the word of one engineer that was there said this guy wasnt there

 

[fazers_on_stun]

Look at this, again AMD reiterates that perf should be inline with i5/i7 SB:
http://www.amd.com/us/products/desktop/processors/amdfx/Pages/amdfx-competitive-comparison.aspx

Something has to be wrong with the benches.

Or wrong with AMD's marketing dept.

That's just a PR blurb - I didn't see any actual performance there. The only AMD performance comparison I saw is when they ran the 8150 against an i7-980x for gaming and an i5-2500K for heavily-threaded workloads (another ).

Wish AMD's pre-earnings quiet period was over so that JF-AMD could explain what happened, as he is much more of an insider than all of us put together..

 

[fazers_on_stun]

You guys should read this, it explains why Bulldozer is a bit of a flop:

"Ex-AMD Engineer Explains Bulldozer Fiasco: Lack of Fine Tuning."
http://www.xbitlabs.com/news/cpu/display/20111013232215_Ex_AMD_Engineer_Explains_Bulldozer_Fiasco.html

IIRC there was something similar a couple months ago on BSN or maybe S/A. Wonder if it's the same guy or a different engineer?

Anyway, if what he says is true, and all those expert engineers who did the hand-tuning of the K8 and up designs have left, then I don't see how AMD could make BD or PD better without rehiring them and spending the $$ and time to fix the problem. Using SoC automated design techniques is faster and cheaper than paying engineers however, so maybe that is how AMD will do things from now on. And it provides one reason why an 'old-school' guy like Dirk Meyer left - maybe he was dead-set against it and thus got canned by the BoD.

If you extrapolate 6-core, 1BN transistor Thuban to 8 full cores, you'd get around 1.4BN transistors, not 2BN like BD has. So that also seems to suggest BD design is not optimized.

 

[fazers_on_stun]

AMD never hyped BD. They described their goals and made ONE announcement 50% for 33%.

Aren't you forgetting that ad where AMD "benched" (as in no system specs provided) a BD against the i7-980X for gaming and against an i5-2500K in multithreaded? That sure seems to suggest cherrypicking..

As the Linux benches come out you'll see what suckers you all are for bowing to a criminal.

Crap. And I finally left all of that. Thanks.

http://en.wikipedia.org/wiki/Advanced_Micro_Devices

It was reported in December 2006 that AMD, along with its main rival in the graphics industry Nvidia, received subpoenas from the Justice Department regarding possible antitrust violations in the graphics card industry, including the act of fixing prices.[5

 

[Chad Boga]

Or wrong with AMD's marketing dept.

That's just a PR blurb - I didn't see any actual performance there. The only AMD performance comparison I saw is when they ran the 8150 against an i7-980x for gaming and an i5-2500K for heavily-threaded workloads (another ).

Wish AMD's pre-earnings quiet period was over so that JF-AMD could explain what happened, as he is much more of an insider than all of us put together..

http://forums.anandtech.com/showpost.php?p=32421412&postcount=106

So here's the story on IPC from my perspective.

The original design goal was higher IPC. Back before we had taped out the first processors the discussion came up. I am not allowed to say anything in public that hasn't been vetted past engineering. I specifically asked the engineering team about IPC and they replied that they expected higher IPC and I was cleared on that statement.

In my estimation, I made the IPC stament on XS and I don't recall making it other places (but I am sure that my comments were reposted.)

This is not a case of me lying, this is a case of me being wrong. And that happens to all of us. There is no way that I would intentionally say something that I did not believe to be true because a.) the interenet is a permanent record and b.) eventually you'll have to answer to those comments.

 

[lp231]

Anyone notice the bulldozer in that tomb stone pic?

 

[fazers_on_stun]

So any thread that casts BD in a positive light gets to keep going?

If certain fanboys including one CR would just mind the TOS, so that the mods wouldn't have to camp out on BD threads, I'm sure those threads could keep on ad infinitum .

 

[fazers_on_stun]

http://forums.anandtech.com/showpost.php?p=32421412&postcount=106

So here's the story on IPC from my perspective.

The original design goal was higher IPC. Back before we had taped out the first processors the discussion came up. I am not allowed to say anything in public that hasn't been vetted past engineering. I specifically asked the engineering team about IPC and they replied that they expected higher IPC and I was cleared on that statement.

In my estimation, I made the IPC stament on XS and I don't recall making it other places (but I am sure that my comments were reposted.)

This is not a case of me lying, this is a case of me being wrong. And that happens to all of us. There is no way that I would intentionally say something that I did not believe to be true because a.) the interenet is a permanent record and b.) eventually you'll have to answer to those comments.

Good find, and according to the rest of his post, he won't be posting in enthusiast forums any more, due to all the hate he's been getting from various fanboys.

A shame that some people apparently cannot control themselves and behave like adults - I for one will miss his posts here. While JF was not an engineer but a marketing director, he provided some insight that we otherwise would not have had. And he always told people here to wait until the product released, look at the reviews, and make up their own minds as to whether it made sense for them to buy BD.

 

[JAYDEEJOHN]

IIRC there was something similar a couple months ago on BSN or maybe S/A. Wonder if it's the same guy or a different engineer?

Anyway, if what he says is true, and all those expert engineers who did the hand-tuning of the K8 and up designs have left, then I don't see how AMD could make BD or PD better without rehiring them and spending the $$ and time to fix the problem. Using SoC automated design techniques is faster and cheaper than paying engineers however, so maybe that is how AMD will do things from now on. And it provides one reason why an 'old-school' guy like Dirk Meyer left - maybe he was dead-set against it and thus got canned by the BoD.

If you extrapolate 6-core, 1BN transistor Thuban to 8 full cores, you'd get around 1.4BN transistors, not 2BN like BD has. So that also seems to suggest BD design is not optimized.

http://www.realworldtech.com/forums/index.cfm?action=detail&id=123260&threadid=123240&roomid=2#CURRENT
redpriest (redpriest@redpriest.something) on 10/14/11 wrote:
---------------------------
>All respect to Cliff Maier, but he wasn't there.

 

[fazers_on_stun]

http://www.realworldtech.com/forums/index.cfm?action=detail&id=123260&threadid=123240&roomid=2#CURRENT
redpriest (redpriest@redpriest.something) on 10/14/11 wrote:
---------------------------
>All respect to Cliff Maier, but he wasn't there.

Seems to be a bit of a 'he said, she said' conundrum then .. But there does seem to be some evidence for a non-optimized design, don't you think? What do those extra transistors in the 8150 do? Besides slow it down, increase the die size and expense, and raise the power consumption that is?

 

[The-Don]

If certain fanboys including one CR would just mind the TOS, so that the mods wouldn't have to camp out on BD threads, I'm sure those threads could keep on ad infinitum .

This.

Once you get to the point that you are relying on insults about another group to make your argument, said argument loses its effectiveness.

The biggest lesson that life teaches is that of humility. Admitting you are wrong is tough. At the same time, it is also empowering. Being able to look back at something and saying "Hmm, I was wrong" removes the need to defend yourself over and over. This "need" to be right seems to overpower rational thinking especially when you are emotionally invested. At some point though, your argument is broken down by the "other side" and you simply can't defend it. It is at this point that insults usually start. The other option is to attack something about the "other side" instead of discussing the real issues.

Unfortunately, this type of behavior is present in all fanboys from both sides. It's ok to be wrong about things, the world won't stop turning.

To the OP: If you are happy with your decision, that is all that matters.

 

[statikregimen]

just remember that amd fanboyism is built on "AMD will get better in the future (and intel will get worse)".



that's not an upgrade.

Meh...It's doing what I wanted it to faster. Is that not the definition of upgrade?

 

[JAYDEEJOHN]

Seems to be a bit of a 'he said, she said' conundrum then .. But there does seem to be some evidence for a non-optimized design, don't you think? What do those extra transistors in the 8150 do? Besides slow it down, increase the die size and expense, and raise the power consumption that is?

My monies on redpriest
Could be northbridge functions/power controls

 

[fazers_on_stun]

My monies on redpriest

Maybe so. You'd think the sites reporting the rumor would look into the guy's credentials, but then maybe not.

Could be northbridge functions/power controls

400M+ extra transistors for NB or power gating? Remember that a 6-core Thuban has about 1 billion transistors - scale it up to 8 full cores gets you 1.33 billion transistors. And IIRC Intel's power gating only cost an extra million transistors.

400M transistors would build around 16,000 Intel 4004's ..

 

[JAYDEEJOHN]

I'll give some speculation
It could be some fusion functioning unit, like how they used the side port, or their early (tho this was pre AMD) tesselation unit

OK, Armadillo hat off now
Maybe is just an unexplained part of the die as of yet, meaning extended layout for certain functions/connects

 

[EXT64]

That's a bummer about JF not posting anymore. Certainly can't blame him, but I'll miss the insight he offered (even if at times it turned out wrong, he never tried to purposefully mislead). And it always amazes me how much people can get worked up by something like this.

 

[JAYDEEJOHN]

Well, its silly season, and JF wasnt given all the infos he was supposed to have, as things changed as time went along.
In the end, those things can change once again, but thats another story

 

[intel4eva]

Meh...It's doing what I wanted it to faster. Is that not the definition of upgrade?

Well according to the toms CPU hierarchy chart, you may not notice a difference unless you upgrade to a cpu at least 3 levels higher?

 

[fazers_on_stun]

I'll give some speculation
It could be some fusion functioning unit, like how they used the side port, or their early (tho this was pre AMD) tesselation unit

OK, Armadillo hat off now
Maybe is just an unexplained part of the die as of yet, meaning extended layout for certain functions/connects

I thought the Trinity part was for fusion..

Anyway, unexplained or unused part of the die? I would think that AMD would have announced extra functionality, unless it currently has none. In which case they should have cut the power traces to it to save juice, improve clocks, etc.

 

[JAYDEEJOHN]

http://semiaccurate.com/forums/showpost.php?p=139356&postcount=18
I do know that anyone trying to pin the blame on Bulldozer performance not using custom design is full of it, though.