What's new
  • Please do not post any links until you have 3 posts as they will automatically be rejected to prevent SPAM. Many words are also blocked due to being used in SPAM Messages. Thanks!

Hardware Canucks: Benchmarkers Guide to the Phenom II

Status
Not open for further replies.

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
<center>
phenomII_logo-1.jpg
</center>
With the onslaught of seemingly endless hardware releases at the conclusion of 08 and beginning of 09, we at Hardware Canucks thought you could use a break from a mile long list of stuffy specifications, features, and benchmark results comparing the Phenom II to other processors. Instead of the typical results pages that go on and on, for the launch of the AMD Phenom II, we will take a bit more un-conventional route to presenting some information about the Phenom II performance. Recent developments in the overclocking and benchmarking world have once again vaulted AMD into the limelight with their Phenom II launch slated for today. News from AMD at recent tech demos has shown that the dreaded cold bug was all but gone allowing operation at liquid nitrogen temperatures of -196C, and possibly beyond. It was this information that we decided to present the Hardware Canucks: Benchmarkers Guide to the Phenom II.

Today we are strictly going to focus on overclocking the AMD Phenom II with three very different mediums. The first of which will be with traditional ambient air cooling utilizing an industry standard heat sink. This cooling medium will provide fully stable 24/7 results as well as the absolute maximum we can squeeze out of the system without getting too crazy with the volts. With these results we will see what the Phenom II will be capable of in pretty much anyone's hands. The next two cooling methods will only be for those that are into benchmarking and willing to go sub-zero or, sub-ambient with temperatures.

The next traditional step in cooling is with water but we will skip over this option and jump straight to a Canadian winter by means of a custom single stage phase change cooler. The phase change we use will not a lot different than that of an Astetek VapoChill or Mach II GT unit that can be bought retail. This setup provides temperatures as cold as -50C but will warm up substantially under load. From there we will take things as cold as the processor can handle with full on liquid nitrogen cooling. As mentioned, this method of cooling provides us with the ability to bring temps down to -196C if the processor/motherboard can handle it.

This should be quite a bit of fun for both us and everyone reading so sit back, buckle up, and enjoy the Hardware Canucks: Benchmarkers Guide to the Phenom II.

<center>
phenomII_logo-2.jpg
</center>
 
Last edited:

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
Phenom II: Mostly New?

<center>
phenomII_mostlynew-1.png
</center>
Gone is the 5 digit naming nightmare and in its place is the three digit naming scheme that is identical to...umm, Intel? AMD has already successfully branded their new Phenom II processors by duplicating the model numbers that Intel has had in place for their new i7 processors for months now. Other models stray from Intel's line-up but the only two processors being released today are the 920 and 940 so we will focus on them. Is this smart? dumb? or just downright cheeky? Perhaps it is all just a coincidence and AMD was completely un-aware of Intel's planned naming scheme...riiiiiight. Here is a look at the taped out processor release as of this point for the upcoming Phenom II processors.

<center><table bgcolor="#666666" border="0" cellpadding="5" cellspacing="1" width="735"><tbody><tr><td align="center" bgcolor="#cc9999" width="100">Model</td><td align="center" bgcolor="#cc9999" width="100">Cores</td><td align="center" bgcolor="#cc9999" width="100">Codename</td><td align="center" bgcolor="#cc9999" width="100">Socket</td><td align="center" bgcolor="#cc9999" width="100">Frequency</td><td align="center" bgcolor="#cc9999" width="335">L3 Cache</td><td align="center" bgcolor="#cc9999" width="335">TDP</td><td align="center" bgcolor="#cc9999" width="335">Release Date</td></tr><tr><td align="center" bgcolor="#ececec" width="100">X4 945</td><td align="center" bgcolor="#ececec" width="100">4</td><td align="center" bgcolor="#ececec" width="100">Deneb</td><td align="center" bgcolor="#ececec" width="100">AM3</td><td align="center" bgcolor="#ececec" width="100">3.0GHz</td><td align="center" bgcolor="#ececec" width="335">6MB</td><td align="center" bgcolor="#ececec" width="335">125W</td><td align="center" bgcolor="#ececec" width="335">Q2 2009</td></tr><tr><td align="center" bgcolor="#cccccc" width="100">X4 940</td><td align="center" bgcolor="#cccccc" width="100">4</td><td align="center" bgcolor="#cccccc" width="100">Deneb</td><td align="center" bgcolor="#cccccc" width="100">AM2+</td><td align="center" bgcolor="#cccccc" width="100">3.0GHz</td><td align="center" bgcolor="#cccccc" width="335">6MB</td><td align="center" bgcolor="#cccccc" width="335">125W</td><td align="center" bgcolor="#cccccc" width="335">01/08/09</td></tr><tr><td align="center" bgcolor="#ececec" width="100">X4 925</td><td align="center" bgcolor="#ececec" width="100">4</td><td align="center" bgcolor="#ececec" width="100">Deneb</td><td align="center" bgcolor="#ececec" width="100">AM3</td><td align="center" bgcolor="#ececec" width="100">2.8GHz</td><td align="center" bgcolor="#ececec" width="335">6MB</td><td align="center" bgcolor="#ececec" width="335">95W</td><td align="center" bgcolor="#ececec" width="335">Feb 09</td></tr><tr><td align="center" bgcolor="#cccccc" width="100">X4 920</td><td align="center" bgcolor="#cccccc" width="100">4</td><td align="center" bgcolor="#cccccc" width="100">Deneb</td><td align="center" bgcolor="#cccccc" width="100">AM2+</td><td align="center" bgcolor="#cccccc" width="100">2.8GHz</td><td align="center" bgcolor="#cccccc" width="335">6MB</td><td align="center" bgcolor="#cccccc" width="335">125W</td><td align="center" bgcolor="#cccccc" width="335">01/08/09</td></tr><tr><td align="center" bgcolor="#ececec" width="100">X4 910</td><td align="center" bgcolor="#ececec" width="100">4</td><td align="center" bgcolor="#ececec" width="100">Deneb</td><td align="center" bgcolor="#ececec" width="100">AM3</td><td align="center" bgcolor="#ececec" width="100">2.6GHz</td><td align="center" bgcolor="#ececec" width="335">6MB</td><td align="center" bgcolor="#ececec" width="335">95W</td><td align="center" bgcolor="#ececec" width="335">Feb 09</td></tr><tr><td align="center" bgcolor="#cccccc" width="100">X4 810</td><td align="center" bgcolor="#cccccc" width="100">4</td><td align="center" bgcolor="#cccccc" width="100">Deneb</td><td align="center" bgcolor="#cccccc" width="100">AM2+</td><td align="center" bgcolor="#cccccc" width="100">2.6GHz</td><td align="center" bgcolor="#cccccc" width="335">4MB</td><td align="center" bgcolor="#cccccc" width="335">95W</td><td align="center" bgcolor="#cccccc" width="335">Feb 09</td></tr><tr><td align="center" bgcolor="#ececec" width="100">X4 805</td><td align="center" bgcolor="#ececec" width="100">4</td><td align="center" bgcolor="#ececec" width="100">Deneb</td><td align="center" bgcolor="#ececec" width="100">AM3</td><td align="center" bgcolor="#ececec" width="100">2.5GHz</td><td align="center" bgcolor="#ececec" width="335">4MB</td><td align="center" bgcolor="#ececec" width="335">95W</td><td align="center" bgcolor="#ececec" width="335">Feb 09</td></tr><tr><td align="center" bgcolor="#cccccc" width="100">X4 805</td><td align="center" bgcolor="#cccccc" width="100">4</td><td align="center" bgcolor="#cccccc" width="100">Deneb</td><td align="center" bgcolor="#cccccc" width="100">AM3</td><td align="center" bgcolor="#cccccc" width="100">2.5GHz</td><td align="center" bgcolor="#cccccc" width="335">4MB</td><td align="center" bgcolor="#cccccc" width="335">95W</td><td align="center" bgcolor="#cccccc" width="335">Feb 09</td></tr><tr><td align="center" bgcolor="#ececec" width="100">X3 720</td><td align="center" bgcolor="#ececec" width="100">3</td><td align="center" bgcolor="#ececec" width="100">Heka</td><td align="center" bgcolor="#ececec" width="100">AM3</td><td align="center" bgcolor="#ececec" width="100">2.8GHz</td><td align="center" bgcolor="#ececec" width="335">6MB</td><td align="center" bgcolor="#ececec" width="335">95W</td><td align="center" bgcolor="#ececec" width="335">Feb 09</td></tr><tr><td align="center" bgcolor="#cccccc" width="100">X3 710</td><td align="center" bgcolor="#cccccc" width="100">3</td><td align="center" bgcolor="#cccccc" width="100">Heka</td><td align="center" bgcolor="#cccccc" width="100">AM3</td><td align="center" bgcolor="#cccccc" width="100">2.6GHz</td><td align="center" bgcolor="#cccccc" width="335">6MB</td><td align="center" bgcolor="#cccccc" width="335">95W</td><td align="center" bgcolor="#cccccc" width="335">Feb 09</td></tr></tbody></table></center>
So what is Phenom II? Well, from what we have pieced together - since AMD wanted nothing to do with this article - the AMD Phenom II 920 and 940 processors are 45nm quad cores with 125W TDP's. Some confusion has people believing they will have 8MB of L3 cache but we can confirm that our samples are all equipped with 6MB of shared L3 cache with each core getting 512KBytes of L2 cache for a total of 2MB. The Phenom II architecture is based on the original Phenom with many refinements to improve clock per clock performance and allow for higher clock speeds. As the chart above outlines, the Phenom II 920 & 940 are the only two processors slated for release today and clocked at 2.8GHz and 3.0GHz respectively. Later versions of the Phenom II will be released in an upcoming "phased" rollout. These versions will include 4MB L3 cache versions as well as three core X3's and varying clock speeds.

<center><table border="0" cellpadding="3" cellspacing="0"><tbody><tr><td align="center">
die_shot-1.jpg
</td><td align="center">
die_shot-2.jpg
</td></tr><tr><td align="center">
die_shot-3.png
</td><td align="center">
die_shot-4.png
</td></tr></tbody></table></center>
The die shot above to the left is one plucked from PC Perspective and labeled as the Phenom II die. Of course, this couldn't be confirmed by AMD or their web site but we did notice a different die shot on the front page of the Opteron section of the AMD site. In their media gallery section, the Opteron die shots are all the same as the Phenom die shots but the front page die is different, it appears to be the same as the one PC Perspective has posted of the Phenom II providing some level of confirmation that it is in fact the Phenom II die.

As today wears on, we anticipate a lot more information to be released about the Phenom II but up to this point, AMD has been AMD and kept very quiet about further enhancements to the Phenom II architecture. As we have mentioned, this article is not sponsored in anyway by AMD so we have no official information directly from the horse's mouth. Everything we have presented thus far has simply been what has been picked up from the wires. We would like to thank Expreview.com for what little information that we have received about Phenom II.
 
Last edited by a moderator:

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
Phenom II: Mostly Similar?

<center>
phenomII_mostlysimilar-1.png
</center>What we haven't talked about yet is how similar the Phenom II is to the first generation Phenom processor. AMD has gone out of their way to not make upgrades a hassle for users and managed to keep the same socket from the first AM2 day's right up to these new Phenom II processors. So despite the fact that the die underneath the integrated heat spreader is completely re-hauled, the package that the motherboard sees is the same. Thus allowing backward compatibility of Phenom II processors with AM2 and AM2+ motherboards, with perhaps a BIOS flash here or there. Going forward, it is unclear just how compatible the - yet to be released - AM3 processors are going to be with AM2+ motherboards. Word on the street is about an extra pin that may prevent that possibility. What is certain, is that the Phenom II 920, and 940 available today are 100% ready to rock in AM2+ motherboards.

<center>
package-2.jpg
</center>
The first photo is of course our Phenom II sample that should be quite familiar to all current AMD owners, on the left is our Phenom II on one side and an X2 5000+ Black Edition on the other. To tell which is which would require some extra sensory ability. They are identical in every possible way. This also makes the package identical to Phenom processors of course. As mentioned, these similarities mean that current AM2+ motherboards are going to be 100% compatible with the new Phenom II processors. In many cases, a BIOS update won't even be required but with some boards, a BIOS update may be needed for any or full functionality. Here are a few charts of motherboard compatibility that we were able to obtain with many other manufacturers having compatible motherboards based on the same chipsets but no press releases announcing a full list.
<center>
mb_com-1.png


mb_com-2.png


mb_com-3.png
</center>
Sources for these lists come from DFI, Gigabyte, and ASUS web sites. You will want to check each respective web site for whether or not your motherboard is compatible or if you will need a BIOS update in order to run a Phenom II. Again, many other manufacturers have compatible motherboards officially supported already, but we simply presented a few of the bigger lists.
 
Last edited by a moderator:

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
Life, The Hardware And Everything

<center>
phenomII_hardware-1.jpg
</center>The center that this article is revolving around is the three different cooling setups we will be testing this processor with. With this type of testing, we will obviously have three different cooling setups. Below are a couple photos of the major hardware players and of course, the complete list of hardware used for the benchmarks we will be looking at shortly.
<center><table cellpadding="0" cellspacing="0" width="90%"><tbody><tr><td align="left">
hardware-1.jpg
</td><td align="left">
hardware-2.jpg
</td><td align="right">
hardware-3.jpg
</td><td align="right">
hardware-4.jpg
</td></tr></tbody></table>
<table bgcolor="#666666" border="0" cellpadding="5" cellspacing="1" width="90%"><tbody><tr><td align="center" bgcolor="#cc9999" width="25%">Motherboard:</td><td align="left" bgcolor="#ececec" width="75%">Gigabyte GA-MA790GP-DS4H</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%">Processor:</td><td align="left" bgcolor="#ececec" width="75%">AMD Phenom II X4 940 ES</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%">Processor Cooling:</td><td align="left" bgcolor="#ececec" width="75%">Thermalright Ultra-120 w/120mm AD1212MS-A73GL 2050RPM/80.5CFM
Chilly1 Single Stage Phase Change
MMouse Rev3. CU w/LN2</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%">North Bridge Cooling:</td><td align="left" bgcolor="#ececec" width="75%">Stock</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%">South Bridge Cooling:</td><td align="left" bgcolor="#ececec" width="75%">Stock</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%">PWM Cooling:</td><td align="left" bgcolor="#ececec" width="75%">Stock</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%">Memory:</td><td align="left" bgcolor="#ececec" width="75%">Team Xtreem 2x1GB PC2-6400 3-3-3 (TXDD1024M800C3)</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%">Power Supply:</td><td align="left" bgcolor="#ececec" width="75%">Ultra X-Pro 750W</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%">Video Card:</td><td align="left" bgcolor="#ececec" width="75%">BFG GTX 280 OCX
GeForce Release 180.48 WHQL</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%">Additional Fans:</td><td align="left" bgcolor="#ececec" width="75%">120mm AD1212MS-A73GL 2050RPM/80.5CFM</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%">Hard Drive:</td><td align="left" bgcolor="#ececec" width="75%">Seagate 7200.9 80GB SATAII 8MB cache</td></tr><tr><td align="center" bgcolor="#cc9999" width="25%">OS:</td><td align="left" bgcolor="#ececec" width="75%">Windows Vista SP1 / Windows XP Pro SP3</td></tr></tbody></table></center>As mentioned, the cooling setups will consist of three different cooling devices, ambient air cooling with a standard Thermalright cooler, phase change cooling by way of a single stage Chilly1 design, and liquid nitrogen via a MMouse Rev3 CU pot. This setup ranges from ambient air temperatures of 23C~24C, to -50C idle temperatures of the phase change, to a potentially -196C of the liquid nitrogen should the processor be willing. Before we can get to any testing, however, let's first take a closet look at the motherboard we will be using today since it is probably the most vital piece of hardware aside from the processor in question.
<center>
mb-1.jpg
mb-2.jpg
</center>The GA-MA790GP-DS4H is based on the AMD 790GX chipset mated to an AMD SB750. This is considered a mainstream motherboard and by no means designed to be a super overclocking monster, it just so happens that it is the motherboard AMD was using at their recent tech demos with the Phenom II so we figured we couldn't go wrong using it with our article here today. The motherboard is only capable of supporting ATI Crossfire so we will only be using a single GTX 280 for all benchmark results. The BIOS supports plenty of voltage for our needs, even without voltage modifications, so this board will be used like it came out of the box. This should bring a lot of credibility to our results here today. With an average price around $160CND, this motherboard should it prove to be a worthy overclocker for this processor, would make for a very nice budget quad core build when mated to the new Phenom II.
<center>
mb-3.jpg
mb-4.jpg
</center>The one big benefit we found with this motherboard before we even powered it up is the fact that it will be dead simple to insulate the motherboard for sub-zero benching. Believe it or not, this can actually be a factor in some peoples purchasing decisions, especially those that will be running a phase change on a 24/7 system. The heat sink setup on the GA-MA790GX-DS4H will also be left stock to truly replicate what the average user can expect from this motherboard with the new processors. We will now take a brief look at the benchmark suite we will be running the setup under for each of the cooling options and then get into each setup with the accompanying results.
 
Last edited:

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
Life, The Methodology And Everything

<center>
phenomII_methodology-1.jpg
</center>This section is a little different from what we would normally be presenting since the entire goal of the overclocking today is not 24/7 stability, instead, it is an afterthought. The entire purpose of today's party is to have a look at what the Phenom II can do on the bleeding edge of stability. In addition to a slightly diverse overclocking methodology, our benchmarking process will be quite different as well.
<center>
setup-1.jpg
</center>Again, our goal is not to compare the Phenom II in a scientific and methodical manner, in fact; we aren't going to compare the Phenom II to anything but itself. Perhaps down the road those type of comparisons will show up on the pages of Hardware Canucks in a motherboard or memory review but not here, not today. As mentioned, we did take a quick look at the 24/7 stability of overclocking on air so let's go over exactly what that entails around here.

As mentioned, this combination of stability tests will only be seen the one time in the Air Cooling section. The rest of the Air Cooling section and only benchmarks that will be seen in the other benchmarking sections are going to be an effective overclockers toolkit of benchmarks.
<center>
b_cpu-z-1.jpg
</center>The ultimate in bleeding edge overclocking, CPU-Z stability is nothing more than that, the ability to get a CPU-Z validation before the system freezes or locks up. CPU-Z is the standard in validating overclocks in the benchmarking community and is required to help validate pretty much every screen shot taken.
<center>
b_pifast-1.jpg
</center>An oldie but a goodie, Hexus PiFast is a single threaded calculation of Pi run in a DOS window. At one time the Hexus version of PiFast was the preferred way of comparing an overclock. The benchmark is still used on HWBot.org for global and hardware points.
<center>
b_spi-1.jpg
</center>The personal favorite benchmark of this reviewer, SPi is another single threaded benchmark that is well on its way to being an obsolete form of benchmark in gauging performance. That said, it is still THE standard amongst overclockers and we have chosen the two standard 1M and 32M calculations for the benchmarks today.
<center>
b_wprime-1.jpg
</center>wPrime is a relative new comer to the benchmarking scene but has already become one of the favorites making its way onto the pages of HWBot.org as a recognized benchmark. wPrime is a multi-threaded application that obviously caters to more cores. Comparing a dual core to a quad core is not a fair comparison at all. Some consider wPrime to be a very valid form of measuring CPU power as it utilizes the entire capabilities of a processor and does not rely on the memory subsystem at all.
<center>
b_am3-1.jpg
</center>Ahh the days of Aquamark 3 as a recognized 3D benchmark. Those days are long gone and AM3 is nothing more than a system benchmark that benefits more from CPU power and the memory sub-system than it does GPU at times. The key to a good AM3 score with a modern video card is with plenty of system horsepower.
<center>
b_01-1.jpg
</center>Not the start of the 3DMark series, but the start of the Futuremark 3DMark series, 3DMark 01 is the sentimental favorite amongst 3D benchers and the first screenshot to hit the forums after a new GPU is released. The one time inconceivable 100K mark in 01 is easily obtained these days but still a goal unachieved by an AMD processor.
<center>
b_03-1.jpg
</center>The first 3DMark to rely almost entirely on the graphics card, 3DMark 03 really lets quad video card setups really shine but still requires solid CPU power to really stretch a video cards legs. These numbers won't be that impressive in this article due to the limitation of a single GPU we find ourselves with.
<center>
b_05-1.jpg
</center>Again, the Futuremark series turned back to benchmark heavily influenced by the system, but still influenced a lot by the GPU. Recent 05 scores have been going through the roof with the introduction of the 4870X2 and high clocks on the Intel E8600 processors. AMD again, hasn't touched a top 05 score in years.
<center>
b_06-1.jpg
</center>Up until this point, the entire 3DMark series has been single threaded but in 06, Futuremark has changed the entire game. Now, in order to compete in 06 a quad core processor is a must have as 06s multi-threaded processor testing eats dual core processors for lunch.
<center>
b_vantage-1.jpg
</center>Continuing in forward progress, 3DMark Vantage is also multi-threaded and also supports PhysX. For our benchmark we will have PhysX disabled as that has been the decided standard amongst benchmarkers due to the in-equality between ATI and NVIDIA cards with it enabled. The CPU power does effect Vantage scores a good bit but not near as much as the GPU so it will be interesting to see what kind of numbers these Phenom II processors put up in Vantage.

As you can see we have a good mix of single-threaded and multi-threaded benchmarks. All of the benchmarks we are going to be running today are counted for points at HWBot.org and it will be easy enough to compare results there with what we have come up with because we have adhered to the guidelines for benching at HWBot.org. I guess that about sums everything up so let's get started with some results. First up, our benchmarking and overclocking on the Thermalright Ultra-120 air cooling.
 
Last edited:

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
The Phenom II At The End Of The Air Cooling

<center>
phenomII_air-1.jpg
</center>Alright, it is finally time to have some fun. Before we get to the good stuff of absolute maximum benching, let's see what our AMD Phenom II 940 engineering sample is capable of for 24/7 operation. Naturally we start off with a couple photos so everyone can see exactly how the setup looked during testing.
<center>
air-1.jpg
air-2.jpg
</center>As we can see, there is one fan pulling on the Ultra-120 and a second fan somewhat pushing, but it is mainly there to get some airflow over the heat pipe setup and the memory. We can also see that we definitely are not using a Thermalright Ultra-120 eXtreme, and just the older Ultra-120. Temperatures may have been a little better with an eXtreme but the Ultra-120 holds its own, even with quad cores. With full disclosure complete, let's see what this setup was capable of for our standard 24/7 stability testing.
<center>click for full size...
</center>After a bit of fiddling with ratios and voltages, we were able to steady the ship and sail straight ahead right around 3750MHz with a HT bus frequency of 2250MHz and a healthy NB clock of 2500MHz. Overclocking the Phenom II is absolutely no different than the first generation Phenom. With the Phenom II 940, the CPU multiplier is un-locked but clocking the HTT provides the ability to adjust multiplier combinations to provide the best combination for performance. Having the NB frequency at 2500MHz and the memory running at DDR2-1000 4-4-4 provides a very nice memory sub-system while keeping voltages relatively low. A few reports had AMD stating 1.55v on air would be safe with these Phenom II processors so we stuck to that, but don't take our word for gospel, we are just going off of what we "heard".

The Gigabyte GA-MA790GP-DS4H really was a breeze to handle. All voltages aside from vDIMM, vCORE, and vNB were left at default and getting the Phenom II to dance was a painless chore, even for someone like myself who has been out of the AMD clocking scene for a little while. While we are looking at a stable air overclock, we might as well take a quick peek at the temperatures this processor is achieving at an overclock of this magnitude. Below is a chart from SpeedFan depicting the average temperatures of the four core readings in red, and the "CPU" temperature in green from a loaded state to idle using Prime95 Blend with the above overclock.
<center>
air-4.png
</center>Ambient temperature was between 21C~23C as measured from directly above the open air setup. As we can see, if these readings are accurate, that temperatures are quite low. The problem is that with stock voltages and frequencies, we see idle temperatures up to 10C below ambient temperature. Notice that at 1.55v we see idle temperatures of 21C in the graph above when we know that ambient is above even that. This means temperature sensors are way off or there are really poorly calibrated idle temperatures at the very least. Based on what we have seen with this processor over the last few weeks. We would venture a guess that the current temperature readings are 10C~15C lower than they should be. Keep this in mind when getting all excited about the temperatures of these new Phenom II processors, for now. With nothing much else to discuss, let's see what we could push this processor to with air cooling. For these benchmarks, we decided that upping the voltage to 1.65v in the BIOS would be acceptable for short runs, so these results are with the slightly elevated voltage from our 24/7 testing as the maximum.
<table align="center" bgcolor="#666666" cellpadding="10" cellspacing="1" width="90%"><tbody><tr><td align="center" bgcolor="#ececec" valign="top" width="50%">CPU-Z Validation
click for full size...
a_cpuz-1.png

CPU-Z Validation</td><td align="center" bgcolor="#ececec" valign="top" width="50%">Hexus PiFast
click for full size...
a_pifast-1.png
</td></tr><tr><td align="center" bgcolor="#ececec" valign="top" width="50%">Super Pi 1M
click for full size...
a_spi1m-1.png
</td><td align="center" bgcolor="#ececec" valign="top" width="50%">Super Pi 32M
click for full size...
a_spi32m-1.png
</td></tr><tr><td align="center" bgcolor="#ececec" valign="top" width="50%">WPrime - 32M
click for full size...
a_w32m-1.png
</td><td align="center" bgcolor="#ececec" valign="top" width="50%">WPrime - 1024M
click for full size...
a_w1024m-1.png
</td></tr><tr><td align="center" bgcolor="#ececec" valign="top" width="50%">Aquamark 3
click for full size...
a_am3-1.png
</td><td align="center" bgcolor="#ececec" valign="top" width="50%">3DMark 01
click for full size...
a_01-1.png
</td></tr><tr><td align="center" bgcolor="#ececec" valign="top" width="50%">3DMark 03
click for full size...
a_03-1.png
</td><td align="center" bgcolor="#ececec" valign="top" width="50%">3DMark 05
click for full size...
a_05-1.png
</td></tr><tr><td align="center" bgcolor="#ececec" valign="top" width="50%">3DMark 06
click for full size...
a_06-1.png
</td><td align="center" bgcolor="#ececec" valign="top" width="50%">3DMark Vantage - Performance w/o PhysX
click for full size...
a_vantage-1.png
</td></tr></tbody></table>The party started relatively high and slowly worked its way down as more cores got involved. Overall the results are pretty shocking for an AMD processor, but at the same time by no means the highest we have seen up to this point for these new Phenom II's. Other users have been posting results that far surpass these, but there are just as many results in about the same range as us. This leads us to believe that without significant changes to the Phenom II process, our results found here today will be about the average that we should expect from retail samples out today. Here is a quick chart to show the scaling of the clocks with the benchmarks ran.
<center>
air-5.png
</center>The air cooling overclocking is a pretty flat chart. At this point in the game, temperatures aren't playing a huge role and CPU voltage is likely the limiting factor. The multi-core benchmarks really hold their own against the single core results. This trend is about to change when we move on to the sub-zero cooling setups.
 
Last edited:

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
The Phenom II At The End Of The Phase Cooling

<center>
phenomII_phase-1.jpg
</center>Moving forward from our air cooling results on to the more exciting aspect of this journey, we turn the cooling position over to a tired old Chilly1 single stage phase change cooler that requires a re-gas and re-tune very badly. The heat output of these new quad core processors when heavily overclocked is no match for this phase change but we will do our best with it anyway.

For those that don't know, a phase change cooler uses a compressor to compress a gas into a liquid that is very cold because it is compressed. This liquid is then fed into a small evaporator that makes direct contact with the processor and cooling it in the process. Phase change coolers can be bought retail from companies like Asetek with their VapoChill and Prometia with their Mach II/GTs. The Chilly1 phase change we are using is a privately built unit that would be very much comparable to a VapoChill phase change in its current state. We will start with some photos of the unit and the move on to the insulation that is required and finally some photos of the setup during testing.
<center>
phase-1.jpg
phase-2.jpg
</center>For those familiar with phase change cooling, there is nothing too spectacular about our unit. It is very basic with just an on/off switch and no control software. It is not designed for 24/7 use and simply used for benching. It was initially tuned for 65nm dual core loads so desperately needs to be tuned for the higher output quad core processors of today. Even still, it should keep temperatures below -20C under load we figure on this Phenom II and that should be plenty to squeeze a good pinch more out of the processor from our air results. As mentioned though, before we get into any sub-zero temperatures, we need to insulate. Preventing condensation from forming is the key to any sub-zero computing.
<center>
phase-3.jpg
phase-4.jpg
phase-5.jpg
phase-6.jpg
</center>To be truthfully honest, it took us about four and a half minutes to go from a bare board to a fully insulated setup. Aside from the single capacitor towards the top edge of the motherboard and the socket, no other cut outs are necessary. A simple rectangle armaflex layer with the PWM capacitors cut out is all this board really needs. We also put a small piece of armaflex tape over the cam end of the socket to ensure a complete seal against the CPU integrated heat spreader since the armaflex layer doesn't cover it. The entire motherboard back side is insulated in our test bench setup so absolutely no warm air will be near the underside of the motherboard causing condensation. Overall we are quite confident that this insulation job will be 100% adequate to do its job, despite the absolute minimum amount of time it took to setup. It isn't often these days that insulating is this easy but it certainly won't get any easier than the GA-MA790GP-DS4H. We will now finish off with a couple photos of the setup in action.
<center>
phase-7.jpg
phase-8.jpg
phase-9.jpg
</center>Our setup today is pretty straight forward and simple, aside from the phase cooler slapped on the CPU. As we can see though there is nothing really fancy with this setup, and only a couple fans providing the cooling for the stock motherboard. It seems about the time that we looked at some numbers and saw what this processor can do at sub-zero temperatures. We will start with the less demanding benchmarks and work our way down to tougher, multi-core, benches.
<table align="center" bgcolor="#666666" cellpadding="10" cellspacing="1" width="90%"><tbody><tr><td align="center" bgcolor="#ececec" valign="top" width="50%">CPU-Z Validation
click for full size...
p_cpuz-1.png

CPU-Z Validation</td><td align="center" bgcolor="#ececec" valign="top" width="50%">Hexus PiFast
click for full size...
p_pifast-1.png
</td></tr><tr><td align="center" bgcolor="#ececec" valign="top" width="50%">Super Pi 1M
click for full size...
p_spi1m-1.png
</td><td align="center" bgcolor="#ececec" valign="top" width="50%">Super Pi 32M
click for full size...
p_spi32m-1.png
</td></tr><tr><td align="center" bgcolor="#ececec" valign="top" width="50%">WPrime - 32M
click for full size...
p_w32m-1.png
</td><td align="center" bgcolor="#ececec" valign="top" width="50%">WPrime - 1024M
click for full size...
p_w1024m-1.png
</td></tr><tr><td align="center" bgcolor="#ececec" valign="top" width="50%">Aquamark 3
click for full size...
p_am3-1.png
</td><td align="center" bgcolor="#ececec" valign="top" width="50%">3DMark 01
click for full size...
p_01-1.png
</td></tr><tr><td align="center" bgcolor="#ececec" valign="top" width="50%">3DMark 03
click for full size...
p_03-1.png
</td><td align="center" bgcolor="#ececec" valign="top" width="50%">3DMark 05
click for full size...
p_05-1.png
</td></tr><tr><td align="center" bgcolor="#ececec" valign="top" width="50%">3DMark 06
click for full size...
p_06-1.png
</td><td align="center" bgcolor="#ececec" valign="top" width="50%">3DMark Vantage - Performance w/o PhysX
click for full size...
p_vantage-1.png
</td></tr></tbody></table>Going from our air cooled results, these numbers are good but not jaw dropping. The strength of the phase change unit accepts the primary blame and shows this when going to multi-core benchmarks. The heat load produced at these voltages with these elevated frequencies is just too much for this phase setup. The WPrime results are the most telling with the difference between WPrime 32M and WPrime1024M being 437MHz. This is because temperatures raise so much after a minute or two of four cores under load. Any more vCORE and wPrime 1024M was into positive temperatures rather quickly. The 3DMark results from 01, 03, 05, and AM3 are again quite nice because they are single core benchmarks that allow the phase change to maintain temperatures of -30C under load. 3DMark 06 faired the best of the multi-core benchmarks but Vantage did quite poorly only being able to squeeze out a run at 4256MHz.
<center>
phase-10.png
</center>The vCORE we imposed for these phase results was 1.70v set in the BIOS. Some of the multi-core benchmarks needed that voltage to be lowered to run stable as heat was causing issues. We have mentioned a couple times that the phase setup we have was idling at -50C and up to that temperature, this system behaved beautifully. We were able to clock the HT and HTT buses without issue or complaint. There are a lot of reports that these chips won't be able to clock either once we go sub-zero but up to this point, our sample seems to be just fine. Let's now see what happens when temperatures are no longer an issue with LN2 on tap.
 
Last edited:

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
The Phenom II At The End Of The LN2 Cooling

<center>
phenomII_ln2-1.jpg
</center>
We will now really put this setup to the test and see how it handles temperatures as cold as -196C. Just the thought of being able to run LN2 all out on AMD is a bit weird for this overclocker. AMD has had cold bug problems for as long as I have been running sub-zero so it has just never been on the menu before. The phase results show this setup can handle -50C no problem so we are certainly optimistic going into this section. As always though, first some photos of the setup during preparation.

<center>
ln2-1.jpg
ln2-2.jpg

ln2-3.jpg
ln2-4.jpg
</center>
Our base layer of insulation remains the same for preparing the motherboard to run even colder than -50C. It held up remarkably well during the phase testing so we are set there. We then simply stack another layer on top of that like we did with the phase change cooler but instead of the evaporator, the base of the CPU pot goes through these layers. We then start with our tube donuts, the hold down, and the top sleeve for the tube. This entire assembly then gets a single wrap of paper towels in case any condensation forms between these layers. The paper towel is actually a heavy duty shop towel and is the standard now for sub-zero insulation thanks to Vince (AKA k|ngp|n). This entire setup then gets another Armaflex sleeve that encloses the entire setup.

Insulation prep for LN2 was as easy as the phase setup and again, we can't do anything but praise the Gigabyte GA-MA790GP-DS4H for its ease of insulation. One other item to mention is how well the motherboard has held up until now. This will be the ultimate test but the Gigabyte board has been flawless in our merciless beating and hope this trend continues. With the pot setup, it is time to bring temperatures down and get to work.

<center>
ln2-5.jpg
ln2-6.jpg
</center>
We can't really see the temperature in the first photo but we just cracked 0C and are heading down. In the second photo we have this great shot of -100C, but unfortunately this was with an un-responsive system running in the background. Somewhere are -80C at the base of the pot, the processor "bugged" out. This is the famous cold bug we have been praying was gone. As it turns out, the cold bug on our processor wasn't too kind to us and to be honest, anything colder than -60C initiated odd occurrences of the system.

Anything from the system freezing at random times to AMD's AOD software causing issues. The only way to reach the clocks we did was to boot at a low multiplier then raise the CPU multiplier within Windows with AOD. Unfortunately, with AOD freezing whenever it felt like, overclocking under LN2 became a tedious battle in futility. Opening CPU-Z, or SpeedFan would sometimes freeze the system after running a 5 minute benchmark where the system was completely stable. Needless to say, our results weren't as "ambitious" or promising as they were on the phase setup. Let's look at those results first, then we will discuss this overclocking battle a little bit more.

<table align="center" bgcolor="#666666" cellpadding="10" cellspacing="1" width="90%"><tr><td align="center" valign="top" bgcolor="#ececec" colspan="2" width="50%">CPU-Z Validation
click for full size...
l_cpuz-1.png

CPU-Z Validation</td></tr><tr><td align="center" valign="top" bgcolor="#ececec" width="50%">Super Pi 1M
click for full size...
l_spi1m-1.png
</td><td align="center" valign="top" bgcolor="#ececec" width="50%">Super Pi 32M
click for full size...
l_spi32m-1.png
</td></tr></table>
Yeah, not quite what we were hoping for. We basically got some SPi work done and a CPU-Z validation then the system went all haywire. We have had subsequent attempts at LN2 cooling on this setup and each time the results are the same. Anything past -60C is just cold bugged and things get all kinds of weird. The known information about clocking Phenom II under cold is that HT link must stay low, below 1000MHz. The other known is that not many chips can clock HTT when at a certain temperature. Our chip is one of those that just can't clock HTT past -50C pretty much. It really seems weird seeing what we were able to do with the CPU under our phase change system.

Clocking HTT or HT were absolutely no problem at -50C, but with a pot temperature of -60C neither want to clock. AMD "users" have been talking in forums lately saying AMDs latest batch of retail chips have solved the low clocking HT when sub-zero. This is great news as 3D performance does suffer with HT below 1000MHz. Not that there is much point, but here is a chart of the clocking like we say with the phase and air cooling.

<center>
ln2-7.png
</center>
We will hopefully be back with some more retail AMD Phenom II processors in the near future with the exact goal of clocking under LN2...with favorable results. Today, however, we have to pack it in and accept that our sample just doesn't like much past our phase change system temperatures.
 
Last edited by a moderator:

3oh6

Well-known member
Joined
Mar 18, 2007
Messages
1,049
Location
Edmonton, AB
So Long, And Thanks For All The Results

<center>
phenomII_conclusion-1.jpg
</center>We initially started writing this article based on the expectations of 6GHz CPU-Z validations and mid to high 5GHz benchmark results...that didn't exactly pan out. What we found out instead, is that even with a seemingly poor sample, the overclocking of the Phenom II appears to be much better than the original Phenom. The "overhead" as AMD is calling it on the Phenom II for air cooling is substantial, and with rather average processor cooling. Stable clocks of 3.6GHz in reasonable ambient temperatures doesn't seem at all out of the question for most users. Of course our statements at this time are strictly opinion based on limited exposure to a couple of early engineering sample processors, but the early results coming to forums from the first retail Phenom II users are starting to back up this data.

Of course, this article covers a lot more than basic air cooling, and was suppose to be focused on unbridled LN2 benchmarking. Unfortunately our experience resulted in some rather weak overclocking results and a very frustrating experience with LN2. Actually, we need to change that to many frustrating experiences with LN2 on our Phenom II samples. There were multiple attempts that all ended in more Window freezes than Apple would even lead you to believe that Vista causes. Despite these issues, we are still somehow optimistic in the idea that retail samples will fight off the cold bug completely like the processors AMD has been touting around at their tech demos. At this point, details on what to fully expect from the Phenom II isn't very clear, but it is starting to look like we - extreme overclockers - will again be hunting for cold bug-free processors like we always have with AMD. Except this time, the chances of finding one will have increased exponentially.
<center>
conclusion-1.jpg
</center>Despite the cloudy haze brought on by a full pot of LN2 boiling off, we did clear some things up today. The Gigabyte GA-MA790GP-DS4H is no slouch when it comes to clocking the Phenom II. Up to -50C under phase change cooling it performed flawlessly in conjunction with the AMD Overdrive software (AOD). Very few motherboards can claim this feat. The BIOS also provides more than enough options to satisfy our tweaking needs, and the voltage options are amazingly adequate, even for heavy sub-zero clocking. This combination of Phenom II and the Gigabyte AM2+ motherboard is extremely easy to get started with. The fact that the Phenom II brings no new wrinkles in how to overclock makes it quite simple for existing users to feel at home within minutes of firing up these new processors.

The fact of the matter is this, in order to upgrade to Phenom II, most AM2+ users will simply need to upgrade the processor. Everyday a list of current compatible motherboards increases, so the rest of your system components can remain untouched. If you are new to overclocking, the Phenom II should be an easy starting point. If you are an experienced overclocker, you'll be running 3.6GHz in a matter of minutes. If you are an extreme benchmarker, the Phenom II brings previously unknown possibilities to the AMD platform with sub-zero cooling. The simple prospect is enough to keep us excited, but we are still prepared to have to dig through piles of processors to get one completely cold bug free...
 
Last edited:
Status
Not open for further replies.
Top