AMD Rolls Out "Warsaw" Opteron CPUs With 12 and 16 Cores

[Marcus Yam]

AMD brings out new server-grade chips with up to 16 cores.

AMD Rolls Out "Warsaw" Opteron CPUs With 12 and 16 Cores : Read more

 

[Stimpack]

Holy crap. "Cores" or not, that's a lot of them!

 

[eklipz330]

i could have dumped their stocks a week ago and make some nice change, but i got greedy... boy do i regret it.

 

[Kai Dowin]

Well, slap me sixteen times and hand me to me mama!

 

[killerclick]

AMD is comically incompetent, I wish they'd die already instead of preserving merely an appearance of competition with Intel.

 

[DRosencraft]

These things will make for absolute computer monsters. Imagine a quad CPU setup with one of these. And only 99W TDP? It's a great day to be a small business/individual in need of workstation power.

 

[Kewlx25]

Piledriver seems to be a lot more efficient when clocked down at these speeds. Kaveri is nearly as fast as Haswell, clock for clock, but it also consumes nearly 2x the power with near 95watt draw during load with 4 "cores". These Warsaw chips have 16 cores, or 4x as many cores, and nearly the same power draw. Warsaw is clocked about 30% slower, but it more than makes up for it with 4x the units.

I could see this competing and makes me interested.

 

[Kai Dowin]

Piledriver seems to be a lot more efficient when clocked down at these speeds. Kaveri is nearly as fast as Haswell, clock for clock, but it also consumes nearly 2x the power with near 95watt draw during load with 4 "cores". These Warsaw chips have 16 cores, or 4x as many cores, and nearly the same power draw. Warsaw is clocked about 30% slower, but it more than makes up for it with 4x the units.

I could see this competing and makes me interested.

Kaveri isn't anywhere close to Haswell clock for clock. In single thread, fixed clock tests, Kaveri's Piledriver cores get a hell of a kicking from Haswell's cores.

 

[antilycus]

It's also not a fair comparison because their is a GPU in Kaveri unlike Haswell (to my knowledge). IF AMD were to die, Intel would quadruple their prices

 

[COLGeek]

This is pure server "gold", particularly when dedicating cores for specific tasks. Of course, these are not intended for game rigs. Comparing to mainstream (desktop) products is of limited value when looking at the needs for which these CPUs are designed.

 

[sykozis]

Haswell is an APU, same as Sandy Bridge, Ivy Bridge, Clarkdale and Lynnfield. It is unfair to compare these processors "clock for clock" though because the architectures are completely different. Bulldozer, Piledriver and Steamroller are designed for heavily threaded environments. Sandy Bridge, Ivy Bridge and Haswell are general purpose processors that perform best on light-threaded tasks.

 

[silverblue]

Haswell has a GPU. In any case, Kai Dowin is correct - until AMD redesigns the architecture, they're not coming close on single threading.

 

[Kewlx25]

Piledriver seems to be a lot more efficient when clocked down at these speeds. Kaveri is nearly as fast as Haswell, clock for clock, but it also consumes nearly 2x the power with near 95watt draw during load with 4 "cores". These Warsaw chips have 16 cores, or 4x as many cores, and nearly the same power draw. Warsaw is clocked about 30% slower, but it more than makes up for it with 4x the units.

I could see this competing and makes me interested.

Kaveri isn't anywhere close to Haswell clock for clock. In single thread, fixed clock tests, Kaveri's Piledriver cores get a hell of a kicking from Haswell's cores.

When only one virtual core is getting used, modern OSes will sleep the other cores, which causes Intel's CPUs to disable hyperthreading, which allows all of the resources of a cpu core to be dedicated to the single thread.

When it came to the multi-core benchmarks, Kaveri was basically tied with Haswell in most benchmarks and only slightly behind in a few others. Almost any game that uses more than one virtual cpu worth of computing will benefit since HT will remain enabled, which puts Kaveri on par with Haswell.

Since we're talking about server CPUs, my guess is this is most of the time. I would like to see some benchmarks comparing the two to see if my theory is even close, like within 20%.

 

[Kai Dowin]

Piledriver seems to be a lot more efficient when clocked down at these speeds. Kaveri is nearly as fast as Haswell, clock for clock, but it also consumes nearly 2x the power with near 95watt draw during load with 4 "cores". These Warsaw chips have 16 cores, or 4x as many cores, and nearly the same power draw. Warsaw is clocked about 30% slower, but it more than makes up for it with 4x the units.

I could see this competing and makes me interested.

Kaveri isn't anywhere close to Haswell clock for clock. In single thread, fixed clock tests, Kaveri's Piledriver cores get a hell of a kicking from Haswell's cores.

When only one virtual core is getting used, modern OSes will sleep the other cores, which causes Intel's CPUs to disable hyperthreading, which allows all of the resources of a cpu core to be dedicated to the single thread.

When it came to the multi-core benchmarks, Kaveri was basically tied with Haswell in most benchmarks and only slightly behind in a few others. Almost any game that uses more than one virtual cpu worth of computing will benefit since HT will remain enabled, which puts Kaveri on par with Haswell.

Since we're talking about server CPUs, my guess is this is most of the time. I would like to see some benchmarks comparing the two to see if my theory is even close, like within 20%.

Hyper threading can be resumed as just being the ability of one physical core dealing with two threads at once for better usage of the pipeline. It barely have any impact in single threaded scenarios, that are where the IPC is measured.

What Haswell (and both current and older AMD and Intel as well) do to improve the single threaded performance when only one core is being used is to boost the clock using the thermal headroom. Basically, a dynamic over clock. Naturally you must know it.

Kaveri tied a dual core Haswell at multi threaded benchmarks. Even though Kaveri's Piledriver modules shares a FPU, they still have the other core physical parts. Speaking clearly, four Piledriver modules (cores) tied two Haswell cores. When benchmarked against a comparable priced Haswell quad core, Kaveri is left behind by a large margin.

That's why I said that Haswell has the upper hand in clock per clock performance. There are countless tests to prove that and it has nothing to do with hyper threading or turbo boost, since those tests are made with fixed clock and only one core (or module) is used on both processors.

-

Also, don't mistake TDP for power consumption. TDP is the measure of how much heat a processor dissipates, in watts. A TDP of 95W does not means that the processor draws just 95W from the wall.

While it's true that it's amazing to pack 16 cores under a 95W TDP, you should notice that AMD do this using the relief from the GCN inside Kaveri, that combines in a 95W TDP package four piledrive modules, 512 GCN "cores" and things such as the memory controller.

Using lower frequencies (and thus lower voltage), AMD uses the thermal room provided by the omission of the GPU to pack so many cores in a 95W TDP chip.

What's great about Warsaw is it's price. I truly don't expect to see 16 piledrivers modules at low clock beating upcoming Xeons that packs 12 higher clocked Haswell cores in most scenarios. The offering for just US$500 is what makes Warsaw interesting.

 

[crisan_tiberiu]

Piledriver seems to be a lot more efficient when clocked down at these speeds. Kaveri is nearly as fast as Haswell, clock for clock, but it also consumes nearly 2x the power with near 95watt draw during load with 4 "cores". These Warsaw chips have 16 cores, or 4x as many cores, and nearly the same power draw. Warsaw is clocked about 30% slower, but it more than makes up for it with 4x the units.I could see this competing and makes me interested.

There is a big difference between thermal design power measured in watts and how much wattage the CPU actually draws under full load. Right now, I'm on a Ivy Bridge laptop with a 3210m. It has a TDP of 35w listed on the ARK page. I am now running it at 100% load on both cores and the total draw is measured at 15.1w by Coretemp. If I then stress test the integrated GPU to 100% load, the value rises to around 28w, still shy of the 35w TDP. The TDP is the absolute limit that the processor can handle before either permanent damage occurs or a small plume of smoke starts emitting from your motherboard in the general area of your CPU. So, generally, the CPU will only ever reach around 80% of the TDP.

my i7 2600k has a 95w TDP rating, when i OC it to 4,4 GHz and push it with intel burn test, core temp shows a max of 130w, so your statement is false.

 

[sykozis]

AMD's Bulldozer and Piledriver based processors have also been proven to draw well beyond their 125w TDP..... The TDP is a rating of the amount of heat that has to be dissipated, presented in watts, and has nothing to do with power consumption at all.

 

[InvalidError]

The TDP is the absolute limit that the processor can handle before either permanent damage occurs or a small plume of smoke starts emitting from your motherboard in the general area of your CPU.

While this is generally true for POWER semiconductors where the TDP is related to the maximum power dissipation within the SOA under a given set of operating parameters and tend to prematurely fail when operated beyond that after all applicable deratings, for chips like CPUs that have tight operating voltage ranges and much lower temperature ceilings, the TDP usually refers to worst-case power draw under otherwise normal operating conditions. CPUs can operate a fair bit beyond their nominal TDP with adequate cooling.

 

[silverblue]

Kai DowinYou keep mentioning modules in place of cores; Bulldozer-family processors contain one or more modules which feature two integer cores and one shared floating point unit. Essentially, for the number of cores stated, divide by two for the number of modules. As such, Kaveri is a two module, four integer core design.The reason for Kaveri performing on a par with the top i3s in multithreaded scenarios is because AMD has increased the width of the front end decoder thus allowing both cores within a module to receive instructions at the same time instead of each alternate cycle. This was the single biggest issue for MT performance. Single threading is hampered by the fact that load cannot be diverted to the other integer core within the module, thus leaving the workload to operate on a small integer core with two ALUs which is half that of a Core-series core. If the module itself was treated as a core then perhaps we'd see an improvement in single threaded performance as the OS wouldn't care whatsoever about the contents of the module. Considering turbo modes affect the module as a whole and not a single core, it's not as if it'd use any more power.

 

[crisan_tiberiu]

Piledriver seems to be a lot more efficient when clocked down at these speeds. Kaveri is nearly as fast as Haswell, clock for clock, but it also consumes nearly 2x the power with near 95watt draw during load with 4 "cores". These Warsaw chips have 16 cores, or 4x as many cores, and nearly the same power draw. Warsaw is clocked about 30% slower, but it more than makes up for it with 4x the units.I could see this competing and makes me interested.

There is a big difference between thermal design power measured in watts and how much wattage the CPU actually draws under full load. Right now, I'm on a Ivy Bridge laptop with a 3210m. It has a TDP of 35w listed on the ARK page. I am now running it at 100% load on both cores and the total draw is measured at 15.1w by Coretemp. If I then stress test the integrated GPU to 100% load, the value rises to around 28w, still shy of the 35w TDP. The TDP is the absolute limit that the processor can handle before either permanent damage occurs or a small plume of smoke starts emitting from your motherboard in the general area of your CPU. So, generally, the CPU will only ever reach around 80% of the TDP.

my i7 2600k has a 95w TDP rating, when i OC it to 4,4 GHz and push it with intel burn test, core temp shows a max of 130w, so your statement is false.

Which statement? The one where I state that there is a difference between TDP and the amount of power a CPU actually uses?

You said that you managed to stress you CPU to a 28w max mesured with core temp, and said that 35w is the absolute limit of you CPU and if your each that number your CPU gonna "blow up". And i said, my i7 2600k has a max TDP of 95W on intel ark page, and i managed to measure 130W (@ 4,4 GHz) with core temp, and it didnt "blow up"

 

[poordirtfarmer]

Ah yes, with lots of DDR3-1600 and a pair of PCI-e SSDs! I wonder how a pair of the 12 cores on a board like the SUPERMICRO MBD-H8DG6-F-O would stack up against a Haswell rig for video editing, rendering … At only $377 each it seems like an interesting option. I’d surely be interested in a review. Hint, hint.

 

[voltagetoe]

I'd love to see how well these score in 3D rendering compared to Intel flagships. And what kinds of numbers would we get in Photoshop ?

 

[K2N hater]

Any AMD CPU is two generations behind intel.

It vastly depends on your software. AMD still reigns loose in terms of cost for highly threaded applications, server scenario. We could use Folding@home and VMWare as fine examples of its usage. In other words, we could also say Intel architecture is best for less threaded applications, such as games - which are poorly optimized pieces of software over a layer of poorly optimized API.

 

[crisan_tiberiu]

Piledriver seems to be a lot more efficient when clocked down at these speeds. Kaveri is nearly as fast as Haswell, clock for clock, but it also consumes nearly 2x the power with near 95watt draw during load with 4 "cores". These Warsaw chips have 16 cores, or 4x as many cores, and nearly the same power draw. Warsaw is clocked about 30% slower, but it more than makes up for it with 4x the units.I could see this competing and makes me interested.

There is a big difference between thermal design power measured in watts and how much wattage the CPU actually draws under full load. Right now, I'm on a Ivy Bridge laptop with a 3210m. It has a TDP of 35w listed on the ARK page. I am now running it at 100% load on both cores and the total draw is measured at 15.1w by Coretemp. If I then stress test the integrated GPU to 100% load, the value rises to around 28w, still shy of the 35w TDP. The TDP is the absolute limit that the processor can handle before either permanent damage occurs or a small plume of smoke starts emitting from your motherboard in the general area of your CPU. So, generally, the CPU will only ever reach around 80% of the TDP.

my i7 2600k has a 95w TDP rating, when i OC it to 4,4 GHz and push it with intel burn test, core temp shows a max of 130w, so your statement is false.

Which statement? The one where I state that there is a difference between TDP and the amount of power a CPU actually uses?

You said that you managed to stress you CPU to a 28w max mesured with core temp, and said that 35w is the absolute limit of you CPU and if your each that number your CPU gonna "blow up". And i said, my i7 2600k has a max TDP of 95W on intel ark page, and i managed to measure 130W (@ 4,4 GHz) with core temp, and it didnt "blow up"

Yeah, because you aren't using the stock Intel cooler. Put the stock cooler on it and crank it up to 130W and get back to me. Anyway, the whole point was that there is a difference between TDP and what a CPU draws. You proved me RIGHT by showing that the actual power used is not the same as the TDP rating higher or lower. BTW, you sound like this:

ok this is my last post on this:
"The TDP is the absolute limit that the processor can handle before either permanent damage occurs or a small plume of smoke starts emitting from your motherboard in the general area of your CPU", what part of ABSOLUTE limit i did not understand? you didnt mention anything about custom cooling, you only said if the CPU reaches the ABSOLUTE TDP rated by the manufacturer it will "burn".
PERIOD

 

[Draven35]

I'd love to see how well these score in 3D rendering compared to Intel flagships. And what kinds of numbers would we get in Photoshop ?

I'll lay odds the answer is, "not as good as Intel."

 

[childofthekorn]

Piledriver seems to be a lot more efficient when clocked down at these speeds. Kaveri is nearly as fast as Haswell, clock for clock, but it also consumes nearly 2x the power with near 95watt draw during load with 4 "cores". These Warsaw chips have 16 cores, or 4x as many cores, and nearly the same power draw. Warsaw is clocked about 30% slower, but it more than makes up for it with 4x the units.I could see this competing and makes me interested.

Kaveri isn't anywhere close to Haswell clock for clock. In single thread, fixed clock tests, Kaveri's Piledriver cores get a hell of a kicking from Haswell's cores.

Kaveri is steamroller, not piledriver. Warsaw is a generation behind.