Another Thuban graces us with its presence this 7th of December – the Phenom II x6 1100T Black Edition. AMD has been churning out a lot of processors recently with some regularity, but they haven’t released a flagship replacement since the x6 1090T in late April.
The best thing about these new releases is their prices. Most every time a new AMD CPU is released, it’s a speed increase over the previous top model at the same price. Being the self-proclaimed price-per-performance champion, that’s straight butter.
These are stock images recycled from the previous x6 1075T review. You can’t have a processor review without some die photos though. With six cores and a slew of cache, these are worth another look.
Plus, they’re prettier to look at than photos of a CPU. Coming in the standard AM3 package, it looks like any other AM3 processor, but with a new and improved model number!
Phenom II™ X6 1100T Specifications
Model Number & Core Frequency: X6 1100T / 3.7GHz (Turbo) / 3.3GHz (Base)
L1 Cache Sizes: 64K of L1 instruction and 64K of L1 data cache per core (768KB total L1 per processor)
L2 Cache Sizes: 512KB of L2 data cache per core (3MB total L2 per processor)
L3 Cache Size: 6MB (shared)
Total Cache (L2+L3): 9MB
Memory Controller Type: Integrated 128-bit wide memory controller *
Memory Controller Speed: Up to 2.0GHz with Dual Dynamic Power Management
Types of Memory Supported: Unregistered DIMMs up to PC2-8500 (DDR2-1066MHz) -AND- PC3-10600 (DDR3-1333MHz)
HyperTransport 3.0 Specification: One 16-bit/16-bit link @ up to 4.0GHz full duplex (2.0GHz x2)
Total Processor-to-System Bandwidth: Up to 37.3GB/s total bandwidth [Up to 21.3 GB/s memory bandwidth (DDR3-1333) + 16.0GB/s (HT3)] Up to 33.1GB/s total bandwidth [Up to 17.1 GB/s memory bandwidth (DDR2-1066) + 16.0GB/s (HT3)]
Packaging: Socket AM3 938-pin organic micro pin grid array (micro-PGA)
Fab location: GLOBALFOUNDARIES Fab 1 module 1 in Dresden, Germany (formerly AMD Fab 36)
Process Technology: 45-nanometer DSL SOI (silicon-on-insulator) technology
Approximate Die Size: 346mm2
Approximate Transistor count: ~904 million
Max TDP: 125 Watts
AMD Codename: “Thuban”
*Note: configurable for dual 64-bit channels for simultaneous read/writes
It’s the same architecture as previous Thubans. In case you were curious, it’s also the same E0 stepping, so no change there. Seems it’s the same Thuban, just with a 100 MHz speed bump over the previous top dog.
Test System, Opponents & Methodology
This system will look familiar to anyone that read the Phenom II x6 1075T review. The only change was the RAM, but it was manually set at the same speed & timings as before.
|Motherboard:||ASUS Crosshair IV Formula|
|CPU:||AMD Phenom II x6 1100T|
|Cooling:||Thermalright Venomous X
with Push-Pull Delta Screamers
|RAM:||G.Skill Flare DDR3-2000|
|GPU:||ASUS Matrix 5870 Platinum|
|PSU:||Cooler Master Silent Pro 600M|
|OS:||Windows 7 x64|
The competition consists of two CPUs from Intel and four from AMD:
|Intel’s Entries||AMD’s Entries|
|i5 655K||Athlon II x4 640|
|i7 870||Phenom II x4 965 BE|
|Phenom II x6 1090T BE|
|Phenom II x6 1075T|
Stock benches were all run three times and the average of those is what you’ll see in the results. Overclocked tests were run once.
For the i7 870, tests were only run at its 24/7 overclock of 4.0GHz. For all of the rest, they were run at stock and their respective 24/7 overclocked speeds.
|Processor||Stock Speed||Overclocked Speed|
|i5 655k||3.2 GHz||4.5 GHz|
|Athlon II x4 640||3.0 GHz||3.74 GHz and 4.0 GHz|
|Phenom II x4 965 BE||3.4 GHz||4.0 GHz|
|Phenom II x6 1090T||3.2 GHz||4.0 GHz|
|Phenom II x6 1075T||3.0 GHz||4.0 GHz|
To make it a fair fight between the two black-edition Thubans and the multiplier-locked 1075T they were all benched with a 4.0 GHz configuration of 267 x 15 with CPU-NB, HT and RAM all set equally.
Please make one important mental note – very few of these graphs’ ranges start at zero. With such high numbers (some in the tens of thousands), this is necessary to be able to discern a graphical difference. So when reading the graphs, take note of the numbers within them as well.
Only one bench was run at stock speed only – the Everest suite of CPU & FPU tests.
The only unexpected discrepancy is Photoworxx, with the 1100T performing noticeably worse than its brethren. While it was consistent in the results, I don’t trust it; there is no logic there. The rest of the tests were as expected, with it out-performing all other AMD processors and the 655K.
Of course, you can see the AES anomaly. From the 1075T review:
The big surprise here is the AES test, where the i5 655K just eclipsed every other competitor. According to Everest, AES is an “integer benchmark [that] measures CPU performance using AES (a.k.a. Rijndael) data encryption.” While Photoworxx and Zlib both are strongly dependent on memory subsystem performance while Queen benefits from the shorter pipeline, it seems AES focuses solely on the CPU’s processing of integers.
I’ll be honest; the thought of replacing the Everest suite with SiSoft Sandra has been on my mind. Anyone who has played with Everest knows it puts up some funky results from time to time. This Photoworxx set of runs may end up being the straw that broke the camel’s back.
Nothing out of the ordinary here. MHz bump = score bump with the 1100T coming out on top throughout.
Overclocking for Stability
Thubans are nothing if not consistent. While individual processors can differ in MHz-per-volt, the three different models of Thuban I’ve benched were all the same: 1.416 Vcore loaded = 4.0 GHz.
On each CPU, I played for a long time trying to find the absolute minimum Vcore necessary to run at 4.0 GHz. Each time, 1.416 V loaded was the result. One tick below on any of the three CPUs crashed while running LinX.
Why 4.0 GHz? Basically, it’s a nice round number and is where all processors since my dearly departed E8400 have run. That overclock generally does everything with plenty of speed while keeping voltages down. It is also a reasonable, attainable speed for the average overclocker to shoot for (with at least a decent air cooler) and running each of them at the same frequency will show if there are any architecture differences or microcode tweaks at play.
Some would (and indeed have) bring up that they came from AMD and have posited they may be cherry picked. That’s all well and good, except the 1090T was a retail CPU, not an AMD-supplied one; so there goes that supposition.
Temperatures were a bit higher than the 1075T at the same speed/Vcore, but that could very well have been the heat sink mount. Admittedly, there was a bit too much thermal paste in there when I pulled the CPU for going cold, so take these temperatures with a grain of salt.
Benchmarking the 1100T
First on the benching block are the 3D benches. Since both 3DMark06 and 3DMark Vantage separate out the CPU scores, we’ll go with those.
3DMark06 and 3DMark Vantage
3DMark06 remains my favorite 3DMark, though 3DMark 11 – which is coming out soon – may just take that title.
The overall score came in third but just barely. When only the CPU score is considered, the 1100T is the top of the heap. With 06, physical cores seem to take precedent over cores + HT.
The 1075T squeezes out a win over the 1100T, though Vantage likes threads just as well as physical cores and that plays out with the i7 870 scores.
As it should, the 1100T comes out on top in both 06 and Vantage when considering stock-clocked scores.
Cinebench R10 and Cinebench R11.5
For some real-world testing, we have the Cinebench pair of rendering tests.
Stock runs beat all others but the 1100T loses to the 1090T and the 1075T in R10 and R11.5, respectively and the 870 in R10 as well. Cinebench R10 can have a bit of variation with multiple runs, so it’s plausible those are all simply margin-of-error differences.
However, R11.5 is very consistent. In three stock runs, the maximum score deviation was 0.01. I’m not sure why there is a difference, but there is one of approximately 4% at the 4.0 GHz level. The 1100T comes out on top at stock throughout.
SuperPi 1M and 32M
Ahh, SuperPi. AMD just doesn’t have ‘it’ when it comes to this bench. Intel just can’t be beat yet (though the coming of Bulldozer has everyone excited).
Note the SuperPi 32M results are graphed in total number of seconds, with the minute:seconds time listed within the graph.
While it might not have the overall ‘it’, the 1100T definitely has ‘it’ relative to every other AMD processor, with the lowest stock and overclocked times of the bunch.
WPrime 32M and WPrime 1024M
Wprime is a completely different story, with AMD competing right up there against Intel. Only the 12-threaded 980X (which costs a kilobuck) shows a large disparity.
The 1100T is the overall AMD winner here as well. What’s interesting is the two extra threads of the 870 offer only a minuscule decrease in times, making these results even more impressive.
Pushing the Ambient Envelope
With the bar set on air cooling on the 1075T, I promtly went about trying to beat the frequencies it obtained.
With an 18 MHz advantage, the 1100T wins by a nose in WPrime. Superpi was somewhat more impressive at almost 100 MHz. For what it’s worth, I have a feeling the 1075T would run around the same frequency, I just failed to push that extra little bit with that review for some reason or another.
One bench for which I did try to push the 1075T was when seeing how high the RAM could get and have a successful Maxxmem run. The 1075T maxed out at DDR3-2110.
Not to be outdone, the 1100T stripped that crown by a hair and produced DDR3-2126. These aren’t at namby-pamby timings either – these speeds on both processors were produced using timings of 7-9-7-24! AMD’s memory controllers have come quite a long way with their Thubans and the 1100T is no exception.
For our final ambient-cooled test, we see how far CPUz would allow us to go.
This one was just barely able to break the 1075T’s already impressive 4.5 GHz by 20 MHz. The best part? It actually validated this time!
Quick note lest anyone think their CPU will produce these numbers all the time – there was absolutely no attempt at stability for the Maxxmem and CPUz runs. These are max frequencies and stability was never even considered.
Now that we have the ability to put our review processors under the ice, there is a fun new section for your reading pleasure! The first chip taken cold was a 1090T for the ASUS Crosshair IV Extreme review. So I mounted up the aptly named ‘giraffe pot’ (thanks to benching team member xoke for that) to see how the 1100T would fare.
The extreme test setup was identical to the ambient setup except for the motherboard, which was switched out for said Crosshair IV Extreme. Hey, why insulate a second board when you already have one ready to go?
Focusing only on 2D for the run (though the 5870 suffered a couple unfortunate acetone-induced battle scars), I started with the hardest to pass – WPrime 1024 – and went easier as the session went on.
Pretty strong times for both benches and definitely impressive clocks, but they didn’t quite get to the 1090T’s results.
Also not a bad SuperPi 1M time and strong clocks. That was also the maximum validation / screenshot it would allow for CPUz.
Unfortunately, just like the 1090T, I couldn’t get a CPUz validation to save my life. Even at the clocks I was booting at (5.0 GHz, even) it would not validate. Infuriating to say the least, but that’s definitely not the CPU’s fault. Someone remind me to write a complaint letter to CanardPC.
So while it didn’t quite give the same results, it was very close and we can’t knock it for that. It only takes a small change in wafer position to account for a few MHz here and there and ever CPU is different, 4.0 GHz results notwithstanding.
Final Thoughts & Conclusion
Everybody likes a new flagship, but the best thing about this one is that it comes without any price increase. Starting out at an MSRP of $265, it’s right where the 1090T was before its recent price drop to $235 MSRP ($229.00 with free shipping at Newegg right now), which was applied leading up to this release.
The 1075T review was quite enthusiastic. It was a new player right in the middle of the lineup where one was definitely needed. Its price has also been reduced by the way, to $199 MSRP, which is where Newegg has it with free shipping. It gave an all-important extra multiplier to a locked CPU.
The 1100T, while still exciting (who doesn’t like a new top-o’-the-heap?), isn’t quite as exciting. It still does what it should do and does it well. With an unlocked multiplier, its strength is in its flexibility. For overclockers, especially the budget-minded overclockers that AMD caters to, it’s tough to see recommending this over the 1090T. Thirty bucks is thirty bucks; and if the 1090T clocks just as well with an unlocked multiplier, why change? If you already own a 1090T or if $30 would break the bank / allow you to upgrade one other component, it may not be the best idea to run out and get one of these.
If you want the absolute best AMD has to offer, go with the 1100T. If you just want to play with a new processor that’s really not very expensive for what it is, do the same. If you plan on ever running at stock clocks, definitely run out and get one of these. It does a superb job at any task thrown at it and will do it faster than any other AMD chip while at least keeping up with, if not surpassing, similarly priced Intel chips.
– Jeremy Vaughan (hokiealumnus)