Why is 945 so much worse??

what actually is the passmark cpu benchmark?

what does it actually test ?

There so many other benches around that show the phenoms to be far beetter performers than intel quads higher up that list . that i suspect the bench is not representative of very much at all

notice the slow doggy Phenom 9600 complete with tlb error rates higher than an e8500 !
WHAT ?

Ok, thats one synthetic benchmark. Do you plan on playing a lot of passmark?
JK. Sorry I couldn't resist.

I would take a look at the cpu charts here on toms, especially looking at the kinds of games you might be playing. Synthetic benchmarks are ok, but they (often) don't tell you much about real world performance.

I would expect most application performance to fall in between the 940 and 955. However those two processors have unlocked multipliers if you were going to attempt any overclocking. If not, then the combo deal may be your best bet.

I also understand that AMD is giving incentives w/ mobo combos with the new X4 965, which may also be worth a look.

Just for the two above, here is what it does:



http://www.passmark.com/products/pt.htm

Its a all around test.

As for why the 945 scores lower than the 940 I cannot say for sure. I really cannot say. i would think a faster CPU would do better. I mean the Intel ones make sense. The QX6800 is a bit below the QX6850 which there is a 70MHz difference and the FSB difference too.

but other than that it makes no sense. Even the X3 unlocked and the X4 910 does better than the 940.

then again the 940 was a DDR2 part and most of the newer ones are AM3/DDR3 parts. So possibly in those tests the DDR3 helps them out?

What Id really like to know about all those bungholio synth tests are, say math in a synth, is that all it does at the time? If so, name a non bungholio anything that does JUST that?
Always, always place real world apps/games ahead of bungholio makrs

Conclusions based only on a single benchmark is ridiculous. Check out also other tests (both synthetic and real app) done for the 945 and see the big picture. I usually treat synthetic benches as second-rate given that no one would use a computer just to run benchmarks.

Ok, so basically don't trust synthetic benchmarks too much then. But still, it's very odd that it has lower performance. There must be some reason.

Besides that, is there any performance difference between the 125W and the 95W versions of the X4 945 on Newegg?

125W - Phenom II X4 945 $169.99 http://www.newegg.com/Product/Product.aspx?Item=N82E168...
95W - Phenom II X4 945 $169.99 http://www.newegg.com/Product/Product.aspx?Item=N82E168...

Any other differences at all in performance etc?

I mean, if there are no differences, then why would anyone purchase the 125W? (Or why would AMD even produce it anymore?)

AMD doesn't produce a 125W version anymore, it's just left over stock newegg is trying to sell. There is no difference other than one is slightly more energy efficient, probably only a few less watts under load because when the Phenom II first launched it barely drew over 95W, but was forced to use a 125W label because of it.

Ahhh, thanks.

One more question, I'm confused about the power chart posted in the recent 965 review:

http://www.tomshardware.com/reviews/phenom-x4-965,2389-...



Is that watts? As in the CPU under load uses almost 600 Watts *alone*??? Meaning you need a 900+ W PSU to run this beast in a normal computer??

If so, then what the heck is the point of putting "140W" on the CPU specifications?

As that graph says, it's AC power going into the PC.

That link doesn't explain their test setup so it's hard to say. I'm guessing that they used high latency DDR 3 for the 945 while they used low latency DDR 1066 for the 940 thus giving the 940 a higher score. In the same AM2+ board and RAM the 945 should score the same as a 940. The 945 does a little better though in an AM3 board with good DDR3 RAM.

Here's a collection of lots of CPU benchmarks: http://www.anandtech.com/bench/default.aspx?p=36&p2=83

I'm not sure I follow?

Let me put it another way. How does the 590 on the graph correlate to the size power supply needed to run this CPU? Does it mean that you need a 600W Power Supply minimum JUST for running the CPU and nothing else?

The chart shows AC power from the wall for the TOTAL system, not just the processor. Look at the test setup as a whole and think of it as you would need 600 minimum for running a similiar system.

No, that ENTIRE SYSTEM drew 590W from the wall. If the PSU was 82% efficient, the system was using 484W of power.
So, it would be running uncomfortably near its rated maximum, but a 500W PSU would have been enough.

Ohhhhhhhhhhhhhhhhh, ok, I see now.  



Lost me again. Where did the 82% come from? And how is 500W enough if it requires 590W?  

The power you see listed for PSU's is there "Max" output, but for real world its more important to know there "Average" output for sustained load. Power supplies are rated by efficiency. The average is around 80-85%, a good PSU will be rated at 85% or higher usually. This means that you take the total output the power supply is listed as, say a 600 watt PSU, and multiply it by its rated percent, so a 600 watt PSU rated at 85% is 510 watts. The 510 is what the PSU is going to handle under constant load. With 600 being the "Max" it can draw, but it wont sustain that load (atleast not for long).

I've been building systems for 15+ years and one thing I've learned, no matter how tight your budget, if you have to skimp on a part NEVER skimp on the PSU...

So why is it 590 * 82%?

Shouldn't it be then 590 / 82% = 720 needed?



Does it waste a lot of extra electricity ($$) if I go overboard? Say for example get a 1000W. Or do PC's never use more than what they actually need?

Converting from AC to DC, as a transformer, yes. Theres never a 1 to 1 scenario

How much waste is there?

Example, if you have a 1000 W PSU and your system uses 600 W, is 400W wasted? Or some fraction of 400 W?

Multiply it, not divide. 590 * 82% (.82) = 483.8

Sorry, not exactly. Assuming the PSU isn't an overrated piece of junk, its wattage is the total amount it can be expected to output (under given conditions, e.g. temperature). Depending on the quality of the PSU, it may or may not be able to sustain this for a long time, or at a high temperature. PSUs are most efficient when they are running at 40%-60% of their rated load. An 80+ PSU will have to be at least 80% efficient even at 100% load. If it is rated for 500W, when running at 100% load, it will pull 500 / 0.80, or 625W from the wall.

A Chokemax PSU loaded to 75% at room temperature will smoke and croak. An Antec SG-650 will put out 700W, 800W at 40C and laugh at you (at 870W, it will politely say "No can do," by quietly turning off, and no harm done).

A 1000 watt PSU is rated to Max at 1000, not constantly pull 1000. The amount of power it actually pull all depends on how much you have running in your system. A bigger PSU may draw more power then a smaller unit that is sufficient for your system but it would be nowhere near the full difference (i.e. in your example no 400 watts would not constantly be wasted).

A 500W PSU that is 80% efficient at 100% load will need 625W from the wall. A 650W PSU that is 85% efficient at the same 500W load will pull just over 588W from the wall; so in this case, the bigger PSU is actually wasting less power than the smaller one.

Even at say half load, if its 500 watts at 80% itll take 270 watts to run and convert the 250 DC watts being used

Ah ok, I get it. So what you're saying is a PSU takes in a certain amount of power but it's not a 100% perfect passthrough to the PC. I didn't get that part. Learn something new every day heh.

So realistically speaking, how much power is wasted from un-utilized capacity? Is the efficiency the *ONLY* factor in this?

Or in other words, say you have a 99% efficient PSU rated at 600 W, and a PC using 500 W. Would only 6 W be wasted?

Oh, and that 37W difference typically becomes heat. That may not seem like a lot, but I doubt you'd be willing to test that by taking a firm grip on the 25W soldering iron I plugged in only ten minutes ago :-).

Oh ok, so if you go overboard on a PSU, you just get a really really hot system then?

The waste is not in un-utilized capacity. The waste is due to limitations on the components and circuits used.
If you're really interested in the inner workings of a PSU, there is some good material at www.hardwaresecrets.com.

The reason I'm asking so many questions is because I'd like to put an order in on newegg for a Phenom II X4 955 and mobo tonight  

And I'm not sure if my little 500 W will handle it (Radeon HD 3850, 1 hd, 1 dvdrw)

So of course, if it's not, I'd get a PSU as well. Hence asking if going overboard was ok to make it future proof (and overclocking!)  

No, if you get a really inefficient PSU, you get a lot of waste heat.

For purposes of the discussion, let's say that RetiredChief has just invented a two THOUSAND Watt PSU that is 100% efficient when operating between 20% and 80% of its rated load. This means that in a system drawing between 400W and 1600W, this PSU will produce NO waste heat.

Consider again the Chokemax PSU, that is 65% efficient with a 200W load. That PSU will be pulling over 307W (200W / 0.65) from the wall, meaning 107W of it is being wasted as heat.

Not just becoming hot, in the case of low quality PSUs, it may even explode.

Based on a GPU chart I saved from an article here at THG, a 3850 needs about 6A max, that's all. If your 500W PSU is of even moderate quality, it will have no trouble with your rig.

A low quality PSU, of ANY wattage, when overloaded, may burn, possibly damaging attached components. Furthermore, the label on a low quality PSU may be an outright lie. I'm baffled as to why the people who knowingly sell them aren't in jail or in the ground. I'm not talking about overzealous optimism, I'm talking about PSUs with internals that aren't rated for more than half of what ends up on the label. When such crap burns, in addition to the special effects, your system may be damaged by spikes and/or electrical noise.
A high quality PSU, when overloaded, will gracefully shut itself off, and nothing will have been damaged.

If you think about it, that would be a form of commercial fraud. I guess nobody, if there's any, brings up this issue to the court which might be the reason why those crooks get away with it. Oh well, out of the topic.  

I think they used a GTX280 in that test - it probably pulled 180w itself.

A Radeon HD3850 pulls around 65w.

Woo! Thanks guys. I just put an order in for some new gear : )

AMD Phenom II X4 955 Black Edition Deneb 3.2GHz Socket AM3 125W Quad-Core Processor Model HDZ955FBGIBOX - Retail
http://www.newegg.com/Product/Product.aspx?Item=N82E168...

GIGABYTE GA-MA790X-UD4P AM3/AM2+/AM2 AMD 790X ATX AMD Motherboard - Retail
http://www.newegg.com/Product/Product.aspx?Item=N82E168...

       



[Edit]: Anyone know what "apparent power" is?