Table of Contents
Just when you think they’re done, AMD ships another Deneb-based chip, saying “hey, we’re not abandoning you just because Bulldozer is coming!” The Phenom II x4 980 Black Edition is a speed bump over the previous Phenom II x4 975 BE, which was itself a speed bump over the 965 BE, which (just one more) was a speed bump over the 955 BE.
We’re certainly not going to argue with speed bumps though. While there isn’t a massive architecture change or a process change, you can’t expect manufacturers to release those multiple times a year. Getting the option for a faster CPU based on the same architecture isn’t something we’re going to argue with!
Speed bumps are a good thing. For overclockers they typically mean they are higher-binned than their predecessors. 100 MHz is 100 MHz, and at 3.7 GHz stock, the 980 BE is the highest-clocked AMD chip released to date.
Same Deneb die, binned higher.
Photos aren’t the strong suit in a CPU review; they tend to look the same. Nothing out of the ordinary to see here except the new model number.
As mentioned above, the x4 980 BE is a speed bump over the x4 975. Clocking in at a very respectable 3.7 GHz stock, this is the fastest Phenom II chip you can find.
Phenom IIT X4 980 ‘Black Edition’ Processor
- Model Number & Core Frequency: X4 980 / 3.7GHz
- OPN: HDZ980FBK4DGM
- L1 Cache Sizes: 64K of L1 instruction and 64K of L1 data cache per core (512KB total L1 per processor)
- L2 Cache Sizes: 512KB of L2 data cache per core (2MB total L2 per processor)
- L3 Cache Size: 6MB (shared)
- Total Cache (L2+L3): 8MB
- 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: 258mm2
- Approximate Transistor count: ~758 million
- Max TDP: 125 Watts
- AMD Codename: “Deneb”
*Note: configurable for dual 64-bit channels for simultaneous read/writes
Aside from the speed bump, all specifications remain the same – Deneb core on the standard 45nm process, 512 KB L1 cache, 2 MB L2 cache and 6 MB L3 cache. Even the price is going to stay the same at ~$195.
Test System, Opponents and Methodology
This test system has seen a few CPUs reviewed on it, so it’s familiar to most of our readers.
|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|
Competing with the Phenom II x4 980 are several varieties of CPU from Intel and AMD.
|Intel’s Entries||AMD’s Entries|
|i5 655K||Phenom II x4 980 Black Edition|
|i7 870||Phenom II x6 1100T Black Edition|
There is definitely some solid competition for AMD’s newest quad-core today. Straight away, I’ll say all but the 655K are more expensive than the x4 980, from $35 for the Phenom II x6 1100T BE to $100 and $120 for the i7 870 and i7 2600K respectively, so please keep that in mind when viewing the results.
In testing, the 980 BE was run both stock and overclocked. Stock benchmarks were run three times and the results were averaged. The graphs you will see are all normalized to the stock 980 BE results. The 980 BE at stock is always 100% and all other results are expressed and graphed as percentages relative to its score/FPS. The actual benchmark results are in parenthesis below the percentage.
Overclocking for Stability
Overclocking the x4 980 Black Edition is fun, as is any Black Edition chip. Bump multiplier, run tests, raise voltage, bump multiplier – rinse & repeat! So, what sort of overclock did we come up with?
4.2 GHz, not too shabby! Loaded Vcore was 1.416 V, as used in testing the two Thubans in their reviews previously. They both managed 4.0 GHz at that Vcore, so this is a solid bump to make up for those two lost cores.
The RAM timings for testing were a little tighter than in that screenshot however. Since these Flare sticks are rated DDR3-2000 / 7-9-7-24, I cranked down on that a bit for testing both stock and overclocked. All tests below were run at DDR3-1600 / 6-7-6-20, as in this SuperPi 32M screenshot.
So, while you won’t see any Deneb chips running DDR3-2000, the IMC is pretty strong for running tight latencies at DDR3-1600. As mentioned, these are the speed and timings at which all tests below were run.
End-users don’t have to stop there though. While there isn’t time to explore every FSB / CPU-NB / HT / CPU multiplier / RAM divider combination for this review, Black Edition chips offer an almost endless ability to tweak and find the sweet spot for your particular system.
As usual, we’ll go with what really matters to most folks first – real-world testing.
Cinebench R10 and R11.5
Cinebench is our rendering test of choice and comes in two iterations. R10 is older and not quite as consistent (which is where averaging three tests at stock comes in handy). R11.5 is extremely consistent, with generally a maximum .01 difference in scores.
Stock results are decent, but overclocked are better. Results scale pretty close to linearly; the 13.5% overclock translated to a 10.3% increase in R10 and 11.7% in R11.5. Against the competition, it didn’t do too bad when you consider the prices mentioned earlier. The 655K is (was) the closest-priced Intel chip and it doesn’t hold a candle to the x4 980.
Everybody compresses and uncompresses files; it’s a fact of life. When you want to make the most out of bakdwidth, compression is the best way to do it. So this one is important for most folks.
The 655K wasn’t tested with 7ip, but results seem comparable to the rendering tests above. All of the compared processors are more expensive, so it stands to reason they will perform better.
3DMark06 and 3DMark Vantage
With today’s hardware, 3DMark06 is almost all CPU. You’ll not see many 3D scores scaling with overclock in overall score like 06 does.
Overclocked, and even at stock this chip doesn’t do too badly at all. Its CPU score is closing in on its Thuban brother and beats the i7 870.
3DMark06 doesn’t take advantage of extra threads quite like Vantage though. It’s not as CPU-bound, but it’s getting more so with each new generation of GPU.
Vantage’s ability to eat up all those threads shows some stronger differences with the six- and eight-threaded competition. The 655K still languishes behind; a full core does not a hyper-thread make.
SuperPi 1M and 32M
Ahh single-threaded SuperPi, AMD’s nemesis since Core2 CPUs. While they perform very well with multi-threaded loads, AMD just doesn’t have ‘it’ for SuperPi.
No surprises here. The tighter memory timings than previously run for AMD testing, as well as the higher frequency definitely helped out here relative to the 1100T. As expected, the Intel offerings just run away with this benchmark.
WPrime 32M and 1024M
WPrime should be a bit better for the AMD chip. Single-threaded performance just doesn’t seem to have been on AMD’s radar for the last couple of years. They believe multiple-thread-optimized CPUs are where it’s at and focus most of their efforts there.
The x4 980 is definitely much better when you throw a multi-threaded load at it. When overclocked it even gets within 15% of the 8-threaded i7 870 and 6-threaded 1100T. Considering the 870 is still a hundred bucks more, that’s not bad.
Max Ambient Overclocks
Pushing overclocks is something we all enjoy, just to see the maximum ability the chip will have. Benchmarkers especially enjoy seeing what happens when you give a chip some voltage. While I was not able to take this chip sub-zero, it’s still enjoyable to see how far it will go on air.
This chip really impressed me with its WPrime and SuperPi ability. No Phenom II AMD chip I’ve toyed with (from a 965 BE to the 1100T) was able to reach 4.7 Ghz for anything on air, much less actually running a benchmark!
It was even able to squeeze a couple MHz more for a CPUz validation. While the validation was rejected (as many high AMD overclocks are), it still shows I was able to boot, overclock and save the validation at an impressive 4779 MHz.
There will be more to come in the future though. As I type this, the CPU is on its way to Phenom II master Dolk (author of the wonderful Phenom II overclocking guide), who will give this CPU a run for its money under liquid nitrogen. We’re definitely looking forward to his results!
Final Thoughts and Conclusion
Let’s first get the award out of the way. This chip gets an Overclockers Approved because it clocks like mad. A perfectly stable, strong 4.2 GHz 24/7 overclock is excellent. 4.7+ GHz maximum clockspeed is even better. So it definitely deserves to be Approved because it overclocks very well and performance is on par with expectations.
It’s great to see AMD continuing their trend of not raising prices on speed bumps. Quoting AMD, “Once again our popular 4-core performance chip gets a speed boost. The new top part will slide into the price point that the current 975 BE is at, and existing models will waterfall down a notch.” The official word is “~$195”. Newegg has the 975 BE selling for $189.99 so I’d expect the 980 BE to come in at the same price point.
That’s a blessing and a curse. The chip isn’t necessarily worth more, so it’s a very good thing they aren’t raising the price. It just calls into question – why this chip and not the 975 BE (Phenom II x4 975 BE review), or even the 965 BE (which is down to a dirt cheap $158.99)? For the average user, that answer is easy – higher guaranteed 24/7 operating frequencies at stock, and likely even higher for 24/7 stable clocks.
For the overclocker, it’s not so easy. Almost everyone with decent cooling can reach 4.0 GHz on a 965 BE. The difference between 4.0 and 4.2 GHz isn’t that great and would hardly be noticeable for most every-day uses. For more frugal overclockers, the 980 BE isn’t necessarily the way to go.
If you already have an AMD Phenom II x4 Black Edition, this chip definitely isn’t for you; wait for Bulldozer. Who this chip will appeal to will depend on several factors. If you have a motherboard that doesn’t necessarily play well with a Thuban (think 700-series chipset boards); or say you have a Phenom II x2, or an Athlon chip and need something that can handle a heavier load, this chip will treat you well.
However, if you are considering or already have an 800-series motherboard that plays well with Thubans, you’ll want to look to an x6 chip. If you have $189.99 to spend on this chip, you’ll likely be able to save or scrape up ten bucks more and grab a Phenom II x6 1090T for $199.99.
So is the chip Overclockers Approved? For its clocking ability, most definitely. Is it worth it vs. another option? That’s up to you!
AUTHOR’S NOTE – We were sent a pricing update after all of the reviews had published. I just posted this in the forums:
AMD just notified us of a price decrease of $10 over what they previously said, so I’m editing the article to reflect the new MSRP of $185.
The good news is this drops the entire processor stack by an additional ten bucks! The bad news for this little CPU – The Phenom II x6 1090T is now the same price, at $185, making that decision a no brainer. If you board can run a Thuban and you want a new CPU, get the Thuban!
It’s definitely something to weigh when considering this particular CPU. It’s still Overclockers Approved because it does its job, does it well and overclocks even better. It’s just unfortunately not as good of a deal as getting the Phenom II x6 1090T at the same price.
– Jeremy Vaughan (hokiealumnus)