Saturday, January 24th 2009

First AMD Benchmarks With DDR3 Memory Posted

Better late than never, AMD is embracing the DDR3 memory standard. Its newest desktop CPU socket, the AM3, connects the processor to dual-channel DDR3 memory. Initial specifications about AMD's processors on the new socket suggest that DDR3 1333MHz (PC3-10600) will be held as the memory standard the integrated memory controllers on AMD's processors support. The far-east team of Tom's Hardware posted the first series of benchmarks of a Phenom II AM3 processor, when paired with DDR3-1333. The benchmarks featured Gigabyte's MA790FXT-UD5P motherboard along with Corsair's dual-channel DDR3-1333 kit running at DRAM timings of 9-9-9-24. The CPU-Z window also shows that the memory modules are running unganged, meaning that the two memory controllers of the Deneb core are independent, with each controller handling a 64-bit wide memory channel. This state is somewhat comparable to that of single-channel memory, except that multi-threaded applications will still be able to independently address memory on each channel, utilizing all the bandwidth on offer.

At the center of the test-bench was the AMD Phenom II X4 910, the company's first flagship quad-core AM3 processor. The X4 910 features a default clock speed of 2.60 GHz. The same chip was tested in two settings: default clock-speeds, and overclocked to 3.12 GHz (240 x 13.0 @ default vCore). The screenshots below show the test results for WPrime. The overclocked chip scored a crunch-time of 46.613 s (1M). The overclock did not affect SuperPi at all, with insignificant differences in the scores between the two. PC Mark 05 got a decent boost, while 3DMark06 didn't. Memory bandwidth and latency tests showed something strange, with the overclocked CPU (in effect memory) turning up with lesser latency (while normally, increase in clock speeds tend to step up latencies). The tests show that there are increments in performance with the use of DDR3 memory, though they are merely proportional to the clock speeds the memory is running at.
Source: Tom's Hardware
Add your own comment

75 Comments on First AMD Benchmarks With DDR3 Memory Posted

#51
eidairaman1
The Exiled Airman
well to get back to your point being, if DDR had a process shrink to say 45nm, its clock could probably be raised without even changing the timings and possibly lower the voltages, it would probably eat the DDR2/3 ram. Another Point being, these benchmarks are synthetic meaning they only Provide a PR number, it doesnt show the true potential of these products, they are also good for testing stability other than COD 4.
LozzaI see what you're trying to say, that now RAM is now more focused on bandwidth instead of latency.

But what I'm saying is that it should be able to produce better results despite this, because if you look at the price difference and how new the tech is you would expect it to be faster than a 2+ year old tech. It's the same as when DDR2 800 came out originally, the figures were similar to DDR 400 because of the latencies. The only thing going for it was the power consumption. That's only for the memory tests though.

But then if you look at Core 2 figures and Phenom figures in a superpi test the Core 2 is better even using the same RAM with the same latencies - so there is still scalability with respect to clock speed/computing power of the chip.

So I would expect the new chips to be better than the old one, even if they are using this "slower" memory.
Posted on Reply
#52
cdawall
where the hell are my stars
blueskynisThat's what I thought too. Is that a normal score?

EDIT: OK, here is the explanation:
www.xtremesystems.org/forums/showpost.php?p=3600818&postcount=3
i already said this in post #3
cdawallthese tests seemed a bit odd when i found them and wprime looks to have been run on a single core when you factor in a 3.89x efficiency factor (about what these get) for the chip it is pretty close to the numbers a phenom 2+DDR2 gives


also this is set for stability not performance so you could easily see them saving several seconds off everything with proper settings
Posted on Reply
#53
Lozza
Yes that is a good point, synthetic benchmarks never do really show the true potential of these things.

EDIT: Ok I've just read that xtreme systems link, it explains it all.
Posted on Reply
#54
imperialreign
WarEagleAUWith the price of DDR3 coming down more and more, its not too much longer till the price vs ddr2 wont really be a factor. Even for the highest latency ram that EastCoast linked us too, its only 51 bucks for a 2gig kit.
I was about to comment on this same thing, but I see ya beat me too it :p


As far as DDR3 prices vs DDR2 prices go - it depends on what kind of, and what performance segment, DDR3 you want to buy.

If you're looking for the el-cheapo, off-the-shelf-at-Best-Buy, not going to OC or hope it runs stable -type of DDR3:

< $25

if you want something cheap, that should be fairly reliable:

< $50

if you're looking for the low-end performance market:

< $100

and the price of more higher-end performance DDR3 is on par with what high-end DDR2 was only a year and a half or so ago.

Even still, one can purchse 2GB of high-end DDR3 at prices comparable to high-end DDR2.


It's time to start putting some of these myths to rest :toast:
Posted on Reply
#55
MilkyWay
whats the point in ddr3 other than fast speeds and power efficiency
the timings are so bad on them

they will get better just like ddr 2 did but right now the speed vs timings is BAD
Posted on Reply
#56
imperialreign
we've been over the timing debate, before.


Just to break it down in a very general way, running timings of CAS8+ with DDR3 doesn't mean squat when the memory is capable of moving nearly twice the amount of information that DDR2 is capable of, per clock cycle.

Think of it like this, DDR2 moves information like a garden hose. DDR3 moves information like a firehose.

Think of timings like kinks in the hose, how many kinks can you add to a gardenhose before it starts to hurt overall performance? How many kinks can you add to a firehose before it hurts overall performance?


DDR2 benefits from getting all those kinks out. Sure, you can increase the volume moving through the hose by increasing the pressure (think: clock speeds), but after a certain point, it just can't do any better without possibly bursting the hose. DDR3, on the otherhand, being a firehose, can handle much more volume and pressure (think: clock speeds) without having to iron out all the kinks (high latency timings).
Posted on Reply
#57
MilkyWay
timings on ddr3 dont matter as much because it makes up for it in raw speed right?

im not on tpu as much due to lack of wireless on main rig, thats why i aint been involved in any recent discussions like ddr3 and phenom 2
Posted on Reply
#58
EastCoasthandle
MilkyWaytimings on ddr3 dont matter as much because it makes up for it in raw speed right?

im not on tpu as much due to lack of wireless on main rig, thats why i aint been involved in any recent discussions like ddr3 and phenom 2
Timing still matters, not so much as it relates to DDR2 but as it relates to DDR3. It's a little confusing but here an example:
People want to know the difference between DDR3-1600 at CL9 vs DDR3-1600 at CL7 when using AM3, for example. You really can't compare DDR2 timing to DDR3 since we know that DDR3 is faster (1600MHz for example). Only a well thought out review can fill in missing pieces of this puzzle. Time will tell (hopefully).
Posted on Reply
#59
eidairaman1
The Exiled Airman
actually pressure and velocity are inversely proportional
as velocity increases, pressure decreases, as pressure increases, velocity decreases

venturi effect

en.wikipedia.org/wiki/Venturi_effect

now with electrons, since apparently they move at speed of light without control, latency and Clock speed manage the flow of electrons thru the system (Local Oscillator), because just like how EM rays are emitted into space, once they hit space (either beyond our atmosphere or within) they are moving at speed of light, once they hit the pickup they are slowed down, aka Radio transmit towers and Radio receivers in cars.
imperialreignwe've been over the timing debate, before.


Just to break it down in a very general way, running timings of CAS8+ with DDR3 doesn't mean squat when the memory is capable of moving nearly twice the amount of information that DDR2 is capable of, per clock cycle.

Think of it like this, DDR2 moves information like a garden hose. DDR3 moves information like a firehose.

Think of timings like kinks in the hose, how many kinks can you add to a gardenhose before it starts to hurt overall performance? How many kinks can you add to a firehose before it hurts overall performance?


DDR2 benefits from getting all those kinks out. Sure, you can increase the volume moving through the hose by increasing the pressure (think: clock speeds), but after a certain point, it just can't do any better without possibly bursting the hose. DDR3, on the otherhand, being a firehose, can handle much more volume and pressure (think: clock speeds) without having to iron out all the kinks (high latency timings).
Posted on Reply
#60
btarunr
Editor & Senior Moderator
eidairaman1actually pressure and velocity are inversely proportional
as velocity increases, pressure decreases, as pressure increases, velocity decreases

venturi effect

en.wikipedia.org/wiki/Venturi_effect

now with electrons, since apparently they move at speed of light without control, latency and Clock speed manage the flow of electrons thru the system (Local Oscillator), because just like how EM rays are emitted into space, once they hit space (either beyond our atmosphere or within) they are moving at speed of light, once they hit the pickup they are slowed down, aka Radio transmit towers and Radio receivers in cars.
lolwut

DRAM latency is brought about by the way the SPD is programmed, on how it makes the IMC/MCH deal with the memory (based on the SKU and quality of the DRAM chips used). higher latency = more number of clock-cycles spent in moving data across. The reason you don't have 800 MHz (1600 DDR) do CL 2T is because of the high clock speed, and the constraints/limitations brought about by the DRAM technology with respect to the chips' voltages and thermal footprints. The high clock speed makes up for the latency, though for PC3-10600 (1333 MHz), the ideal latency should be 6-6-6-~18. The latest generation of DDR3 kits haven't managed to do those latencies at sub-1.6V, owing to the silicon fab technology they're built on, and hence we're seeing 1333 MHz kits with 9-9-9-~24, which isn't all that great compared to 1066 MHz at 5-5-5-~15.
Posted on Reply
#61
frankie827
isnt that a bad 32m score?
i get 13.638sec seconds with my setup

i know i have 4gb of ram while that setup only had 2gb...but still....
Posted on Reply
#62
imperialreign
eidairaman1actually pressure and velocity are inversely proportional
as velocity increases, pressure decreases, as pressure increases, velocity decreases

venturi effect

en.wikipedia.org/wiki/Venturi_effect

now with electrons, since apparently they move at speed of light without control, latency and Clock speed manage the flow of electrons thru the system (Local Oscillator), because just like how EM rays are emitted into space, once they hit space (either beyond our atmosphere or within) they are moving at speed of light, once they hit the pickup they are slowed down, aka Radio transmit towers and Radio receivers in cars.
thanks for spoiling my fun :p :laugh:

yeah, I know all that physics stuff - but I was trying to keep it on a technical level most would be able to understand :toast:
Posted on Reply
#64
unclewebb
ThrottleStop & RealTemp Author
Intel better watch out. AMD has almost caught up to the CPUs that Intel was turning out 2 years ago. :laugh:
Posted on Reply
#65
eidairaman1
The Exiled Airman
unclewebbIntel better watch out. AMD has almost caught up to the CPUs that Intel was turning out 2 years ago. :laugh:
cheers :nutkick:
Posted on Reply
#66
kid41212003
unclewebbIntel better watch out. AMD has almost caught up to the CPUs that Intel was turning out 2 years ago. :laugh:
:laugh:
btarunrDRAM latency is brought about by the way the SPD is programmed, on how it makes the IMC/MCH deal with the memory (based on the SKU and quality of the DRAM chips used). higher latency = more number of clock-cycles spent in moving data across. The reason you don't have 800 MHz (1600 DDR) do CL 2T is because of the high clock speed, and the constraints/limitations brought about by the DRAM technology with respect to the chips' voltages and thermal footprints. The high clock speed makes up for the latency, though for PC3-10600 (1333 MHz), the ideal latency should be 6-6-6-~18. The latest generation of DDR3 kits haven't managed to do those latencies at sub-1.6V, owing to the silicon fab technology they're built on, and hence we're seeing 1333 MHz kits with 9-9-9-~24, which isn't all that great compared to 1066 MHz at 5-5-5-~15.
Pretty much sum up what I wanted to say.
Posted on Reply
#67
TheScavenger
unclewebbIntel better watch out. AMD has almost caught up to the CPUs that Intel was turning out 2 years ago. :laugh:
Regardless, it's good to see AMD moving in a positive direction :)
Posted on Reply
#68
eidairaman1
The Exiled Airman
finally someone else that sees what they are doing instead of bashing them like the majority do here. I think the only reason they do it is so they can mask their insecurities.
Posted on Reply
#69
TheScavenger
eidairaman1finally someone else that sees what they are doing instead of bashing them like the majority do here.
I think a lot of the AMD-bashers fail to consider the price of the CPU (or entire system) when comparing an AMD processor to an Intel. I understand that Intel has been wiping the floor with AMD in terms of maximum, high-end performance but none of that matters when you're on a budget. Its hard to go wrong with an X2 7750 BE when you can't spend more than $75 on a processor.

Josh
Posted on Reply
#70
eidairaman1
The Exiled Airman
TheScavengerI think a lot of the AMD-bashers fail to consider the price of the CPU (or entire system) when comparing an AMD processor to an Intel. I understand that Intel has been wiping the floor with AMD in terms of maximum, high-end performance but none of that matters when you're on a budget. Its hard to go wrong with an X2 7750 BE when you can't spend more than $75 on a processor.

Josh
the Kuma is a Considerable Jump compared to what Brisbane/ Phenom 1 brought to the table, too bad they wont have a BE/FX in that niche anymore.
Posted on Reply
#71
cdawall
where the hell are my stars
frankie827isnt that a bad 32m score?
i get 13.638sec seconds with my setup

i know i have 4gb of ram while that setup only had 2gb...but still....
it was only run on a single core
TheScavengerI think a lot of the AMD-bashers fail to consider the price of the CPU (or entire system) when comparing an AMD processor to an Intel. I understand that Intel has been wiping the floor with AMD in terms of maximum, high-end performance but none of that matters when you're on a budget. Its hard to go wrong with an X2 7750 BE when you can't spend more than $75 on a processor.

Josh
i already proved that with that cpu in everything except superpi it beat C2D clock for clock
Posted on Reply
#74
frankie827
cdawallit was only run on a single core
not trying to start a fight here but, if it was run on a single core, why is cpu-z recognizing all 4 cores. and dont say that it recognizes them even if they are disabled because i know for a fact that it wont.
Posted on Reply
#75
Meizuman
frankie827not trying to start a fight here but, if it was run on a single core, why is cpu-z recognizing all 4 cores. and dont say that it recognizes them even if they are disabled because i know for a fact that it wont.
In Wprime 1.55 you have to manually set the thread count each time you open the program.

With one thread my system did 46,562s.
CPU 3.2GHz, NB/HT 2.4GHz, RAM 1066 5-5-5-15
Posted on Reply
Add your own comment
Dec 22nd, 2024 19:57 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts