5000+ Brisbane or 5600-6000 Windsor

Ok looking to upgrade my single core Orleans 3800+ that runs at 2.91GHz

It's down to:

"Black Edition" Brisbane
5600+ Windsor
6000+ Windsor
FX-62

Now my issues are:

Gotta Flash the BIOS for the Brisbane
6000+ and FX-62 are 125W
I only have a 520W PS, but it's not a gaming rig!
I want to be over 3.0GHz with the dual core
I run 4 sticks of 1GB RAM so this will slow me down some.

The Brisbane will use the least amount of power so better for my electric bill and PS.

The 6000+ and FX-62 obviously are killer processors that need a much better chip fan/heatsink. I do have a whole H2O setup but no time to get it up and running yet. Doesn't say I wouldn't for one of these.....

Is the 5600+ going to be the best comprimise of Wattage (89w), ease of install and performance? It's also the cost leader for a bolt in CPU since I don't have to plunk down $65 for a Ultra-120. In terms of just processor price it's $130-$160, the $30 ain't gonna kill me here.

Really am I going to run into a PS issue here with only 520W?

Does getting the FX-62 or 6000+ just give me a big enough kick in the *** to get my H2O setup running? What does your average H2O version of one of these run at? 3.3-3.5+Ghz? For that I would do it.

The computer is really just a work computer that is used for heavy calculations work and running my bussiness. I also watch DVR recordings while I do e-mail etc... so it's a multitasker as well.

What would you all do in this instance?

Don't get the FX-62, its an older stepping and won't clock as well. The 5600/6000+ are both F3 Windsor steppings and should clock to 3.2-3.5 pretty easily with proper cooling. The 5000+ BE should do the same range. However with less cache and high latency cache it will perform slower at the same clock speed (like 5-10% max).

Wattage really isn't an issue, all these chips will use similar amounts of power when clocked up and pumped with volts.

520W is actually plenty even for a high end gaming rig (single video card) provided the power supply is a high quality one that is actually capable of outputting its rated wattage. For example a Corsair 520 could run a heavily oced quadcore and an 8800GTX without issues, but I wouldn't trust a Raidmax 520 with a stock A64 single core and integrated graphics.

I'd either go with the 6000+ or the 5000+ black edition.

6000+ has F3 and double the cache, which can do 3.2~3.4ghz and a relatively high multi, which si good since you are already stressing your NB with 4 gigs of ram. Your other choice would be the 5000+ BE because it has the unlocked multiplier and can prolly also hit 3.2ghz.

I'd still lean towards the 6000+ in the end, just for double the cache.

Well the 5600+ will hit 3.2 as well and also has 2x1MB cache, so why pay extra?

funnyperson1 said:

Well the 5600+ will hit 3.2 as well and also has 2x1MB cache, so why pay extra?

Click to expand...

higher multi, less stress on NB.

Ok so sounds like we are looking at the 5000+ BE and 6000+ Windsor.... Windsor sounds like the performance winner to me. Basically the computer runs the equivalent to SuperPi for hours on end, and the faster the better for getting my results back to me.

The 6000+ is the winner there is what it sounds like?

More thoughts are great to hear from you guys on this...

BTW the PS is a OCZ Modstream 520W that plugs along nicely in now it's second computer. Previously it was a 939 Opty 146, and i built this AM2 rig to upgrade later... which is now!

How about the ligitimacy of this egg review....

"Pros: Very fast, but at the price of comfort and piece of mind...or at least an ultra hard-core cooling solution.

Cons: I'm amazed I'm giving this CPU a bad rating because I'm a die-hard AMD fan. But this chip sucks and someone's got to come out and say it. Consider that the X2 5600+ runs at 93% the speed of this chip, but only 2/3 as hot. This chip consumes 125 watts as opposed to the 5600+'s 89 watts. My 6000+ is running idle in my Shuttle XPC at 50-55C. That's right, idle (with Artic Silver Ceramique no less)! An almost-as-fast 5600+ would run idle in my exact same system at about 35-39C, so think about that carefully before you buy this thing. It's a fast chip but you will pay a major heat penalty.

Other Thoughts: AMD is a great chip maker but they should abandon designs like this that are dramatically different then their siblings. Hapless buyers like me can get duped into thinking its just a faster version of the 5600+. I wish I had done a bit more homework and bought the 5600+ instead."

Go with the X2 6000+. I have that in my rig at work with the stock hs/f in a case with only 2 120mm @1300rpm fans and psu with a 120mm fan and i havent seen load temps above 45c in a hot office. If you spend some good money on a nice air cooler you can oc that sucker with no problems. If you have water cooling you could put that on and drop your temps and maybe get a better oc out of it than on air.

SStrokerAce said:

How about the ligitimacy of this egg review....

"Pros: Very fast, but at the price of comfort and piece of mind...or at least an ultra hard-core cooling solution.

Cons: I'm amazed I'm giving this CPU a bad rating because I'm a die-hard AMD fan. But this chip sucks and someone's got to come out and say it. Consider that the X2 5600+ runs at 93% the speed of this chip, but only 2/3 as hot. This chip consumes 125 watts as opposed to the 5600+'s 89 watts. My 6000+ is running idle in my Shuttle XPC at 50-55C. That's right, idle (with Artic Silver Ceramique no less)! An almost-as-fast 5600+ would run idle in my exact same system at about 35-39C, so think about that carefully before you buy this thing. It's a fast chip but you will pay a major heat penalty.

Other Thoughts: AMD is a great chip maker but they should abandon designs like this that are dramatically different then their siblings. Hapless buyers like me can get duped into thinking its just a faster version of the 5600+. I wish I had done a bit more homework and bought the 5600+ instead."

Click to expand...

They are still talking about stock speeds. Once you overclock power usage by both chips should be nearly identical. The 5600+ and the 6000+ are really the same chip, its just that at the factory someone decided to stick one with a 14X multiplier and the other with a 15X multiplier and a higher stock voltage. Once you start increasing clock speed and voltage, AMD's rated TDPs go out the window.

higher multi, less stress on NB.

Click to expand...

I understand we aren't talking about a huge price difference here, but c'mon, what kind of pos board could someone be running that could do 215HTT, but couldn't do 230HTT to reach 3.2Ghz. Not to mention that he already has his board running at ~245HTT with his current chip.

funnyperson1 said:

I understand we aren't talking about a huge price difference here, but c'mon, what kind of pos board could someone be running that could do 215HTT, but couldn't do 230HTT to reach 3.2Ghz. Not to mention that he already has his board running at ~245HTT with his current chip.

Click to expand...

Didn't notice multipliers are already so high for these AMD chips. I've been used to having a decision between 8x,9x,10x multis all my life.

venice 3000+ = 9x
3800+ x2 = 10x
E2140 = 8x

wen it comes down to 14x and 15x, u are totally right, no difference.

White_Pawn said:

higher multi, less stress on NB.

Click to expand...

I actually lowered my multi to x13 when i overclocked (256htt), as i see alot doing it for some reason...

Also the 6000 is 125w and the 5600 is 85w = cooler, right? or am i wrong?

funnyperson1 said:

They are still talking about stock speeds. Once you overclock power usage by both chips should be nearly identical. The 5600+ and the 6000+ are really the same chip, its just that at the factory someone decided to stick one with a 14X multiplier and the other with a 15X multiplier and a higher stock voltage. Once you start increasing clock speed and voltage, AMD's rated TDPs go out the window.

I understand we aren't talking about a huge price difference here, but c'mon, what kind of pos board could someone be running that could do 215HTT, but couldn't do 230HTT to reach 3.2Ghz. Not to mention that he already has his board running at ~245HTT with his current chip.

Click to expand...

Ok, good info there.... so with the H2O setup and the RAM I have now theoretically I could get 3.675GHz out of the 6000+ and 3.43GHz out of the 5600+. I know I could easily run into problems before that but it does give me a ceiling either way. On H2O I would like to see 3.3-3.5GHz when I'm beating on it...... EDIT I realize the record holders out there are stable in the 3.4-3.5GHZ range so it is something big to shoot for.

One more thing here.... if the power output is not a problem since if you push any of these chips it's going to use the same power output wouldn't the 5000+ Brisbane be just as legit of a option as the 6000+ since the multi is unlocked but your penalty is the smaller cache?

Well the Brisbane IS a different chip made using different manufacturing methods and its power usage will be different from the other chips when overclocked (would probably be slightly lower at the same clock speed).

The main knock against the Brisbane is that its slower clock/clock. Even if it does do 3.4, it still isn't faster than a Windsor at 3.25.

I think 3.4 is a nice goal to shoot for with any of these chips, with 3.2 realistically being almost guaranteed.

The power consumption really isn't a big issue, at similar clock speeds the power consumption difference is going to be like 5-15W max.

If you really are concerned about power consumption, get a new high efficiency power supply. In any case the Modstream is a mediocre psu, and you would do well to replace it. Compared to the old, inefficient Modstream design at like 75% efficiency, a newer 80+ psu would significantly lower your power usage, and allow you to get a higher quality psu that will last you a long time. This one has gotten excellent reviews from reviewers who actually push the power supply to its rated wattage and open it up to check build quality:
http://www.newegg.com/Product/Product.aspx?Item=N82E16817139003

White_Pawn said:

Didn't notice multipliers are already so high for these AMD chips. I've been used to having a decision between 8x,9x,10x multis all my life.

venice 3000+ = 9x
3800+ x2 = 10x
E2140 = 8x

wen it comes down to 14x and 15x, u are totally right, no difference.

Click to expand...

NB is on board and runs at CPU clock so it's all about the CPU speed as multi's are just for tuning. You do have to watch HT still as that is off die.

funnyperson1 said:

The main knock against the Brisbane is that its slower clock/clock. Even if it does do 3.4, it still isn't faster than a Windsor at 3.25.

I think 3.4 is a nice goal to shoot for with any of these chips, with 3.2 realistically being almost guaranteed.

Click to expand...

Explain that one..... slower clock/clock? Even with the higher GHz the computer actually calculates at a lower speed?

SStrokerAce said:

Explain that one..... slower clock/clock? Even with the higher GHz the computer actually calculates at a lower speed?

Click to expand...

The brisbane has a slightly slower cache (high-latency cache) and thus needs more time to access it. That means that data is accessed slower, then maybe processed as fast as a windsor at the same GHZ, but is, again, written back to the cache slower.
This can be quite a penalty if you process lots of little information, for example, as switch times are quite huge.

It does only show a noticeable difference in certain applications, though and most of the time the difference will not climb above 3%.

I think the app with the greatest impact is super pi.

It is something to consider, anyways..

If you care for power efficiency, take the advice with the PSU, it made a huge difference for me (and the air quality in my room) Also look that you get an adequate wattage PSU, because it's high-efficiency wattage is usually at a fixed point, and anywhere above or below that can take efficiency down to like 60% even on a 80+% efficiency PSU..

The cache latency and amount differences between the Brisbane and Windsor are negligible in real-world use. You're looking at maybe a couple tenths of a second slower in the SuperPi 1M test. Meanwhile, the Brisbane will also be drawing less power and outputting much less heat. The Brisbane will also be less expensive.

If you're looking for ultimate speed, then I think the 5000 Black Label is your best choice if you're going to overclock it. If you're not going to overclock (yeah, right... why would you have asked a bunch of overclockers for advice ) then the 6000 is the best choice.

If you're looking for high speeds with practically imperceptible speed differences clock for clock, while using less power and producing less heat AND being the least expensive choice, then I say go with the Brisbane.

Ok so that explains the L2 Cache and it's impact on calculation speeds. I wondered why my old Opty 146 did so well in SuperPi compared to my Orleans 3800+ when I had a 200MHz gap between the two. Also could easily be the RAM impact as well.

Yes I'm planning to overclock whatever I get, the big task this computer has other than e-mail, surfing etc... is LONG calculations. What I use to stress test the computer is what it runs on, combustion engine analyisis software. I can get higher heat temps for the durations it runs, sometimes up to 18-24hours straight with this than running with most of the stressing software out there. Since it's just calculating data based on mathematical forumlas SuperPi performance is what I look at the closest. 3% difference can easily mean a 1/2hour longer I have to wait for results, 5% would be another hour!

I realize that at this point Intel's stuff beats AMD at SuperPi, but I'm a AMD fan based on a few things, one being that they were the first on F1 cars years ago with Ferrari and have sponsored local events here like the NASCAR race at the Glen.

The Brisbane got me interested in my upgrade to a new chip initially, but I have to research all the choices out there that are comparable. The 6000+ looks like it could be better, I mean it does have a 15X multi how much more can I need? I've looked at the 6400+ and that might be able to get me my 3.5GHz for my work on H2O but it's still at a higher price premium compared to the other two models.

Thank you for the help so far with this....KEEP IT COMING!!! I hate the "pick this for me" posts in forums, but I really don't know nearly as much about chips as I do about engines so I have to ask, ask, ask to feel like I make the correct choice here.

In light of your scenario, I'd go with either the 5000 or 6000. The price gap really isn't that big, so perhaps the 6000 really is your best choice. Personally, I'd grab the 5000 and pocket the difference.

Your Opty outperformed the 3800+ because SuperPi is only single-threaded which means it will only use one core, and the Opty has twice the cache which helps out in retrieving- storing-sending-retrieving-storing-sending operations. Same thing for the Brisbane-Windsor debate. Ultimately, the Windsors will be faster and if you're looking at long-term projects over periods of 24 hours, then yes, that tiny cache difference will translate into a half hour or so of your real-world time, and that is significant. A half hour can be the difference between making or missing deadlines.

In your case, I like the 5000 but the 6000 is probably your best bet.