The AMD Phenom faced a fairly rocky start, but the guys in green are trucking forward and have started releasing new processors into the wild. These new chips are of a different stepping, and feature fixes for many of the identified issues. Today were are looking at such a processor, checking to see if these early problems have been truly been fixed.
For those that have checked out our original review of the Phenom 9600 Black Edition, the major issues we outlined was the TLB Erratum fix that hindered performance, and the mediocre overclocking ability for a factory unlocked processor. For information on the Phenom processor, and AMD’s Spider platform. I suggest checking out our 9600 Black Edition review for the beans.
As for this review, we are focusing purely on those problems, if they’ve been corrected, how they’ve been corrected, and how they affect performance. Here’s a quick recap of what makes the Phenom processor such an interesting product. Full specs can of course be found on the product home page.
- The industry’s first true Quad core x86 processor
- AMD64 with Direct Connect Architecture
- AMD Balanced Smart Cache
- AMD Wide Floating Point Accelerator
- HyperTransport™ technology
- Integrated DDR2 DRAM Controller with AMD Memory Optimizer Technology
- AMD Virtualization™ (AMD-V™) With Rapid Virtualization Indexing
- AMD Cool’n’Quiet™ 2.0 technology
- AMD CoolCore™ Technology
- Dual Dynamic Power Management™
When it comes to pricing, AMD hasn’t quite come out ahead. Our review processor rings in at $209 USD street price. The closest competitor from Intel – the Core 2 Quad Q6600 – rings in at the same price. That being said, the Q6600 is 100MHz slower. The Core 2 Quad Q9300, which matches the Phenom clock for clock, is $269 USD. We’ll have to see how this will translate in the long run, but first we’re going to cut right to the chase and start benchmarking.
The Test Rig
To test this latest processor from AMD I simply installed it in my standard test rig, replacing the Phenom 9600 BE that previously resided in the socket. The rest of the specifications for the test rig read as follows:
- Asus M3A32-MVP Deluxe/WiFi-AP 790FX Motherboard
- OCZ Platinum XTC REV.2 PC2-6400 2GB 2X1GB DDR2-800 Memory Kit
- ZOTAC GeForce GTX 280 Video Card
- Western Digital 640GB Caviar SE16 7200rpm SATA2 Hard Drive
- Thermaltake DuOrb CPU Cooler
- Gelid GC1 High Performance Thermal Compound
- NZXT Tempest Midtower ATX Case
- Thermaltake Toughpower 1000W Power Supply
To perform the tests, we ran a myriad of benchmarks with the CPU set in various modes. We had the Phenom set to TLB Enabled, TLB Disabled, and then we tried to overclock this unlocked beast as far as it would go. As for software, we chose from a variety of synthetic benchmarks, multimedia tests, and real-world gaming runs. All that’s left is to see how those benchmarks pan out, and how good a job AMD has done at correcting the issues with the Phenom.
Real-Time Price and Stock Check – Check More Products Here
The Tests
Once the Phenom was installed and the test system was set back up with Windows Vista, I set myself to the task of testing everything out. First and foremost, I want to run some simple synthetic benchmarks to see if the TLB erratun had truly been fixed, and if that fix had incurred a performance hit. Before we do that though, let me elaborate on exactly what the TLB erratum is.
Prior to the Phenom’s initial release, it was discovered that a bug in the translation lookaside buffer (otherwise known as a TLB) in AMD’s quad-core processors could cause a system lockup in rare circumstances. This of course included the first Phenoms, which were of the stepping codes “B2” and “BA”. A few BIOS and software workarounds appeared to prevent this problem from occurring by disabling the TLB, but these workarounds typically incurred a performance hit of at least 10%. That has since been corrected in hardware. Processors with the bug fixed have “50” as the last two digits of the model number, with stepping “B3”. The processor we’re reviewing here falls into such a category.
3DMark Vantage
The first benchmark we ran to test these different modes was a quick run through on 3DMark Vantage in Performance mode. Though this latest version of 3DMark is meant to test video cards, it does have a CPU component in it. To test the different environments in which the CPU would operate, I went into the BIOS and made to appropriate changes. OverDrive is a little too flaky for my tastes, and would hard lock the system when I was just changing the TLB mode. Anyhow, I shot for a modest overclock for these tests. I was able to push the Phenom 9850 to 2.8GHz with only having to bump the Vcore to 1.325V. I then hurried up and waited for each test run to complete.
As you can see, the TLB fix being enabled or disabled has no effect on the overall performance of the Phenom 9850. The processor works to it’s full potential all the time. As for the overclocking numbers, they are definitely a significant jump over stock speeds. The CPU test tracks almost linearly, producing a 10% performance increase with the 12% increase in processor speed.
PCMark05
When we switch to the PCMark suite of tests, things get a little more muddled. PCMark05 has a more clear cut definition of CPU performance in its testing, which is the reason we used it. The main composite score generated showed some minor variances, with disabling the TLB fix garnering a minor amount of additional performance over it being enabled. Overclocking the processor of course gave the best performance numbers, with the 300MHz jump in CPU speed producing between a 4% and 9% increase in performance. This doesn’t quite match the 12% jump in speed, but this is just the composite score, and also factors in video card and memory performance.
Isolating the benchmark to just the CPU tests is where the full story really starts to unfold. The CPU tests showed the expected outcome, with little to no difference between the TLB fix being enabled and disabled, and overclocking performance producing some significantly higher test results. These results also scale better, with the overclocking of the Phenom 9850 producing an 11% performance increase.
Real-Time Price and Stock Check – Check More Products Here
PCMark Vantage
Unfortunately results in PCMark Vantage, the latest test suite from Futuremark, weren’t as clear cut. Rather than a series of tests designed to torture a particular subsystem, the new PCMark tests the system in 7 different tasks, generates a composite score for each test, and using some complex maths, generates a final score from that. So for all that trouble, it’s no surprise that the final score generated shows little variance between the different modes of operation we setup on the Phenom 9850.
Trying to find any single test that has a heavy weighting towards processor performance was a little tricky. That being said, the tests for rendering and encoding video (TV and Movies) and the productivity tests made the most sense to use for focused CPU testing. Performance differences in the Productivity test were negligible, but the TV and Movies test produced a 4% performance increase when overclocked. This isn’t really anything to write home about. Looks like we’ll have to manually encode some video and audio to really test this processor.
Multimedia Testing
Futuremark’s benchmarking suites give a decent representation of system performance, but sometimes a much more precise instrument is needed to test a particular component in a computer. In the case of a CPU like the Phenom 9850, video and audio encoding are as good a test as any.
For video encoding we used Nero Recode 2.0. This video encoder supports many of the different portable formats the average user may use to re-encode their favourite content, and it’s fully multithreaded. This allows it to push any processor to maximum, showing how well it performs in pure processing. We chose the iPod conversion due to the overwhelming popularity of Apple’s little white MP3 player.
The movie of choice needed to be something long with a lot of detailed action and bright scenes. I chose Star Wars: Episode II – Attack of the Clones for this task. The results from that were in line with what we’ve seen so far, and a little easier to decipher than the PCMark Vantage scoring. It goes without saying that overclocked performance was better then stock speeds. However many reviews have shown that the Phenom 9850 performs at par with, or better then its Intel counterparts in media encoding. The performance jump of 14% from just a 12% overclock does fall in line with that.
Moving from video encoding to audio compression, we used the LAME MP3 encoder and Exact Audio Copy to compress a full length CD into V0 quality MP3s. Our CD of choice for this test was Echoes, Silence, Patience, and Grace from Foo Fighters. EAC was set to use 4 threads in the hope of putting 100% stress on all four cores of the Phenom 9850, and it did for a very short amount of time. Audio encoding really isn’t the processor whore it used to be, and the tests performed at both stock and overclocked speeds took less then a minute, with the overclocked run of course taking less time.
Real-Time Price and Stock Check – Check More Products Here
Gaming Benchmarks
Now the fun begins, and it’s for a reason better than simply being able to play computer games because I “have to”. Gaming benchmarks use real world performance to show whether or not a processor like the Phenom 9850 has any affect on performance, and how drastic that affect is. We weren’t able to balance FPS and RTS games as well as in the Phenom 9600 review, but the lineup chosen makes for a good cross section of popular games and decent benchmark titles.
Supreme Commander
Certain games don’t age well as benchmark titles, which is the reason why we dropped Company of Heroes for this review. The performance shown in that game just doesn’t scale. It appears that Supreme Commander is first mate on that same boat. The graph shown here is rather deceiving, as the performance difference between stock speeds and overclocked was less than 1% when running the built-in benchmark at maximum settings with a resolution of 1680×1050. Though the benchmark for SupCom is another composite score, the CPU is weighted a fair degree due to the amount of AI present in the game.
So with all that in mind, it’s no surprise when I say that gameplay was smooth and without issue. While playing, I ran Fraps to see what framerates were being produced. They never dropped below 60 frames per second. I think we’ll need to move to some more punishing games. Bring on the 3D shooters.
Half Life 2: Episode 2
The first 3D shooter to be tested was Half Life2, and specifically Episode 2. This latest instalment takes place in the backwoods around City 17, and has some additional graphical tricks over the original Half Life 2. For testing we used the same HardwareOC benchmark tool we used in our review of the ZOTAC video card. The tool was set to defaults, with the resolution ramped up to 1680×1050, and we ran through both included demos. Each demo showed a 6% and 7% performance difference, respectively, between stock speeds and overclocked speeds.
Playability was a non-issue with this and most other games based on the Source Engine. The defaults used in the benchmark tool are fairly low, and I can still average 70FPS when ramping up settings and adding antialiasing and anisotropic filtering. With the processor overclocked, that average rises to the high 70s.
Real-Time Price and Stock Check – Check More Products Here
Call of Duty 4: Modern Warfare
From EP2 I moved on to a 3D shooter I thought would present a much greater challenge to the Phenom 9850. Call of Duty 4: Modern Warfare is a graphically rich game that succeeds in its simplicity, and unlike EA’s attempts it actually works. For testing, we ran through the first part of the first USMC mission and using Fraps recorded the average framerate. This produced some very good numbers. Overclocking the processor produced a 10% increase in performance over stock.
Graphically, game play held up well. Our benchmark level contains a lot of close areas, which drives up framerates. But when I got out to more open terrain, or night scenes with lots of lighting effects, I was still easily pulling 60 – 70 FPS average. The overclock helped a bit, with an increase in performance of between 9% and 11% depending on AI and physics load.
Crysis
The final game, and the final benchmark, consists of none other then the butcher of performance known as Crysis. For testing we once again turned to the folks HardwareOC, and their benchmarking tools. The Crysis tool comes with separate demos used for testing CPU performance, which go by the names of Ice and Island. Ice is obviously the more punishing of the two, as consists of a timedemo of our technologically enhanced soldier laying waste to a small village using a big gun. With objects flying everywhere and explosions abound, this test puts pressure on the CPU for physics and the GPU for graphics. Island consists of some destruction, and then throws in some wlaking around in the jungle. This makes it less punishing, but still puts a good deal of stress on components.
Actually the two demos are rather telling of the two sides of playing through Crysis with my current hardware. It’s typically a smooth experience, but when action really heats up graphics rendering can start to lag. And I’m not talking a minor fire fight with a few enemies; I’m talking about levelling small villages with large weapons. Still the performance of the Phenom 9850 was admirable. I think it’s time we wrapped this up.
Real-Time Price and Stock Check – Check More Products Here
Final Thoughts and Conclusion
I really have to hand it to AMD for listening to the community, stepping up to the plate, and making the changes required to make their latest product perform the way its expected to. Like I said in my review of the Phenom 9600 Black Edition, this is one of the things about AMD that has won over many of their now biggest fans. Unlike that previous review, there isn’t any cold hard truth to be faced. Though this version of the Phenom still doesn’t quite meet up with its Intel counterpart, it’s definitely an improvement.
In our test, the video and audio processing were no brainers in that the Phenom 9850 was just as faster or faster than a similarly clocked Intel chip. Overclocking was much easier, much more stable, and actually produced a decent performance gain. Even game performance was nothing to sneeze at, and was of course helped by the ability of this chip to overclock.
The one major downside that the Phenom 9850 carries is that it still isn’t as fast as a similarily clocked Intel chip, as many reviews have shown. As for our review, we also had problems pushing the processor beyond the level of overclock we reached. It’s not that the chip wouldn’t do it; it’s that it got too hot for my comfort. I would recommend a very powerful cooler if you wish to overclock this, and really any, processor.
In the end, the AMD Phenom 9850 Black Edition is a decent product at its price point. The performance isn’t up there with the big dogs, but were talking only by a minor percentage. Really the performance doesn’t need to be an exact match to Intel, because the price is right. AMD’s parts still ring in cheaper then many Intel equivalents. That coupled with the performance seen in this review make this a good product for the overclocker on a budget who wants to personally reign in as much performance out of their rig as possible.
The Good
- TLB erratum fixed
- Smooth overclocking
- Good multimedia encoding performance
- Cheaper then equivalent Intel part
The Bad
- Close, but still behind Intel in gaming performance
- High temps when overclocked
Overall Rating: 8.5/10.0
Discuss This Review in the Futurelooks Community Forums
Real-Time Price and Stock Check – Check More Products Here