Slashdot Mirror
Athlon 64 3400+ Reviewed
SpinnerBait writes "Unlike the Athlon 64 FX-51, this new
3400+ rated Processor, has a 64 bit memory interface, with its integrated memory
controller, drops in at several hundred dollars less than an FX-51 and is also
clocked at 2.2GHz. It gives a P4 3.2GHz Canterwood based machine a run for
its money too, as
this review with benchmarks at HotHardware reports. And where is
Prescott? Fortunately for AMD, it's a bit tardy to market and this will give this new Athlon 64 speed bin time to take a firm hold."
http://anandtech.com/cpu/showdoc.html?i=1941
"I drank what?" - Socrates
Anandtech
Looks like a winner to me!
Mod parent down. The tarrato link will redirect you to goatse.cx
Don't anthropomorphize computers, they don't like it.
http://amdmb.com/#News-7458 or linkified.
Reviewed by amdmb, HotHardware, Neoseeker, CPU Performance, Tech Report, Hardcoreware, Hardocp, Hexus, X-Bit Labs.
To-do List: Receive telemarketing call during a tornado warning. Check.
Shouldn't bee too long with Sun's announced support of AMD chips. They already have it for the Itanic. http://java.sun.com/j2se/1.4.2/install-linux-64.ht ml
Why is this moderated as 'funny' and not 'ignorant' :-P ?
Processor makers 'bin' processors. That is, they try for the fastest speed, but if the chip doesn't make it, it get's 'binned' dowm the line and tried as a lower-speed chip. They can also 'bin' due to market-reasons (putting hi-grade chips in the low-speed bin because of demand, etc)
Guys we have all the reviews listed on our main page, and I'm adding more as they come in. It currently totals at 19. Does Hothardware pay Slashdot for these links? ;)
ignorance is bliss. googlefiberatx.com
Also does anyone have an idea how expensive the AMD 3400+ chips are? Because the AMD 3200+ chips are $400 retail. The article quoted a price for a thousand quantities but I was wondering how much it would cost for just one. Because if its pricey enough the P4 3.2 may beat out the 3400+ dollar for dollar.
Though Intel doesn't have to really worry about that title. At $164 the Pentium P4C smokes the pants off any AMD processor in its price range. At least, after overclocking it to 3 GHz, which is very doable even with standard cooling.
Any peice of hardware that can spank the competition EVEN while its potential isn't fully being realized by the software testing it deserves my dollar.
And yes, im talking about how well it games, I can really give a flying fsck about how quickly it runs office...
What's a speed bin?
In case you're not trolling, chip manufacturers crank out one design of chip, test it, then put them into bins based on how fast they can run reliably. They probably don't actually use plastic bins, but you get the idea.
Thus, a "speed bin" - a lot of chips designated to run at a certain speed, despite the fact that it's the same design and metal as a chip designated to run at a slower speed.
Actually, Apple's in a good position here. Mach allows "fat" binaries (ie, more than one binary in a single application icon). So both version can be distributed together.
They did a similar thing around the transition to PowerPC.
I have a shitty sig!
Due to minor process variations (random faults in materials, equipment, background radiation etc.) every manufactured chip is different. Some chips work fine at higher speeds, some chips only work properly at lower speeds, some chips fail to work at all. Since these microprocessors are at the cutting edge of silicon process technology, the variation matters. Now, to sort out which chip works at which clock speed, the manufacturer has to test every chip and classify them accordingly. Some are sold as 2 GHz chips, some are sold as 1.8 GHz chips, and so on. These different grades are called "speed bins".
Code that has been optimized for the G5 by simple re-compilation will run without penalty on a G4. If you have done more in-depth, G5-specific tuning (levels 1, 2 and 3) then you will in all likelihood want to provide a separate binary. In extreme cases, you may decide that you need only offer one version of your software that runs on Power Mac G5 computers only. However, you'll probably want to support most or all of the Macintosh product line, which means that you need to decide how best to deliver the right code to each of your customers. There are several ways to achieve this; the first is:
Create different versions of your software for each processor that you support. This requires that you maintain three parallel code bases, something you may not want to do.
It is possible for your software to query the computer on which it is running to see which processor-related features are available. You can design your software to isolate processor-dependent code and call the appropriate version as needed. This leads to two additional strategies for packaging your application:
For every function that calls processor-dependent binary code, have your code call the appropriate version. If such functions are needed frequently, using this approach may decrease execution speed and make your source code (cluttered with if...then constructs) less readable.
Isolate processor-specific functions into frameworks or shared libraries, then have your software load the appropriate version when it starts up. This enables you to write your main code without wrapping function calls in if...then constructs."
(from G5 Optimization)
You are missing a little bit. There's a reason why AMD started going with giving their processors numbers instead of clock ratings, mainly because clock ratings are starting to mean less and less.
Clock ratio is only one of the things that is indicative of a CPU's power... you might want to consider comparing the 64 and 32 bit varieties of this chip somewhat akin to comparing a 1Ghz P4 Vs an AMD, or a 266Mhz P1 vs a P2-266.
You may also want to take a look at this review at our good old Tech Report.
That's what "ignorant" means. From latin, ignorare, "to be unaware [of sth], not to know". Antonym of scire ,"to know, to be aware [of sth]". (cf. "Science").
I don't have the numbers in front of me right now, but at a guess you would need 300 watts to cool a 100 watt CPU, so would need to dissiapte 400 watts to air.
It is inadvisable to make any attempt to get the chip below zero, obviously ice formation will happen, and when you switch off, it will melt. Should you switch on again, disaster is quite probable, unless the PCB had a good conformal coating and the socket has an interfacial seal. The conformal coating can be dealt with quite easily, but I have never seen a sealed CPU socket. BTW I usually work as an avionics designer, where we have to make things that will run from well below zero to well above, so I do know the problems.
Another issue is thermal fatigue. The temperature coefficient of expansion of silicon does not exactly match that of the (probably epoxy) package, every temperature cycle causes a stress cycle, which causes a strain cycle, until something breaks. Same for the motherboard itself of course, if you should cool the whole thing. That is also a good reason to never overclock anything, apart from the possibility of getting subtle data errors and increasingly buggy OS as a result of inadequate timing margins, you will definitely wear the thing out a lot quicker. Every 10 deg C roughly halves the life, or the number of on/off cycles it will survive, and if you do the calculations, the numbers are quite depressing for a modern PC.
If you really want a thumping great 64 bit processor (I certainly do, when the price comes down!), it would be best to calculate the cooling system, and maybe do some tests with thermocouples etc, to try to get the CPU chip to settle down at a relatively safe temperature, say 40 deg C, without getting ice formation on the coldest parts. The clever bit would be to get it to power on and off without any excursions below room temperature (often 20 deg C) or above 40 deg C. Heat soak when you switch off the CPU would be minimal, the mass of the chip itself is very small, but cold soak from a huge peltier block could be a problem, the CPU could be dragged down to -40 deg when you switch off, which is exactly what you don't need, for a long and reliable life.
The other thing to watch out for is that at low temperature the CPU internals will be out of spec. It is actually possible to get excessive current flow in some transistors, and local hot spots, because it is too cold. There may also be timing problems, data corruption, ...... It is not possible to test properly that these things are not happening. It takes a smallish time to fully exercise an 8-bit processor to verify all possible data and instruction operations, but somewhere near the lifetime of the universe to do the same for 16 bits. Throw in onboard cache, 64 bits, etc, and it is just impossible. These things work statistically, AMD know how timing variations, for example, might vary across the chip, and allow sufficient margin, within the published clock frequency and temperature range, but deviate from these ratings and this is no longer true. If running at 1GHz, a 1 in 10e12 error rate would corrupt your data or OS code within 1000 seconds, just over 1/4 hour. The error rate required to run an OS for weeks at a time at several GHz defies all attempts at testing. Again, a very good deterrent to overclocking (BTW my non-overclocked ancient K6-350 had a meltdown due to fan failure, and as it died the corrupt Win XP blew away all the passwords so when I got a new CPU and fan, any of th
Blackdown has released J2SE 1.4.2 for AMD64 a few weeks ago, see http://www.blackdown.org/java-linux/java2-status/j dk1.4-status.html.
You fanboys just dont get it. 32 bit vs 64 bit has nothing to do with speed. The 32 bit mode that Mathematica runs in is not the "slower" mode. All of the performance increases in Athlon64 are due to architectural enhancements that are completely independant of the size of the registers...
disclaimer: I am no expert on CPU stuff, I just do a lot of math computations.
If he's actually doing any serious work in Mathematica then 64 bit does start to matter. High powered math is one of the areas where having 64 bit word sizes can make a significant difference. Of course, that's not making much difference in terms of speed, but rather precision (unless of course you're having to do lots of nasty juggling in 32bit mode to handle the required precision).
Jedidiah
Craft Beer Programming T-shirts
There is a HUGE demand for these desktop chips. AMD has pretty much sold out of them.
Your point about people not really needing these processors is valid, but the demand is there.