Phenom vs Ryzen: Hexa-cores a decade apart (Part 6)
Hexa-cores vs dual-core
With the back story out of the way, we will be taking a look at few benchmarks with the Phenom II X2 550 included. Note that the 550 wasn't retested and the performance numbers show here for 550 were taken much earlier with older version of Windows 10 and GPU driver. So, this isn't exactly an apples to apples comparison. Still we march ahead, for science!
Overclocked results for Phenom II X2 550 is also included. Back in those days, you could get sizable increase in clock speed by overclocking in contrast to almost negligible overclocking headroom in modern AMD processors. AMD's Precision Boost Overdrive(PBO) automatically overclocks your CPU if the cooler has enough thermal headroom and the system has power to spare. It does a fine good job too, better than manual tuning in most cases.
To overclock the 550, higher clock speeds were achieved by increasing the FSB and voltage as the multiplier was locked, thanks to it being a non-black edition part. A maximum overclock to 3.72 GHz was achieved from base clock speed of 3.1 GHz, stable enough to complete a Geekbench 4 run but not enough to get through 30 minutes of Prime95 stress test loop. Interestingly, scores were actually lower than stock score! Fine tuning and adjusting the Vcore and FSB, ultimately the highest I was able to get was 3.64 GHz with a positive impact on score and also passing the 30m Prime95 test (though sadly, not the 1 hour loop). An impressive 17% overclock with the stock cooler! We shall see the results with this overclock along with stock results in the select few benchmarks presented in this section.
Fun fact- if you won the silicon lottery, you might have been able to unlock 2 extra cores turning your dual core 550 processor into a quad-core one. Magic, right? Well, not quite. Silicon wafer yields are never a 100%. The best chips with no defects and satisfying (or exceeding) performance targets are sold as the higher-tier products. This process of selectively categorizing chips based on performance results is called binning. Due to the nuances of CPU fabrication process and its complexity, not all chips perform the same even with same features. Depending on you luck, your chip might just clock a bit higher or be better in some other metric. This is informally called "silicon lottery". Now, instead of discarding chips with some defects, some of their defective features are disabled and sold as lower tier chips.
With the Phenom II X2 550, some of the chips were rebranded by disabling 2 cores with potential defects. Now if you were luck enough, you might be able to unlock both those cores and achieve stable results, effectively turning it into a quad core processor (showed up on BIOS as a hybrid processor named Phenom X4 B50) or, maybe even a single core, effectively turning it into a triple-core processor. Unfortunately, mine wasn't so lucky and unlocking any of the cores resulted in instant BSOD (Blue Screen of Death) error. Upping the voltage didn't help either. That concludes the tale of potential free core upgrade.
Now that we have completed the back story of the back story (no more back stories I promise!), lets dive into the benchmarks.
The single-threaded score of 5600X was almost double the multi-threaded score of 550 at stock, let alone the multi-threaded score. So lets ignore the bully for a moment.
550 when overclocked, improved upon its single-threaded and multi-threaded scores by 16% and 17% respectively. 1055T was still faster than an overclocked 550 by 105% in multi-threaded score but saw 550 pull ahead from its 11% deficit in single-threaded score to beat it by 5% after overclock.
In 3DMark 11 Fire Strike results, we also have data for MSI GeForce GTX 1050Ti Gaming Edition running on the same system with 1055T, albeit with older Windows version and graphics drivers. 1055T beat the score of 550 by 19% but that seems insignificant compared to the 37% better score 5600X achieved with Radeon R7 260X over 1055T or, a massive 63% better score than 550. With 5600X, R7 260X really closed the gap from 52% lower score (paired with 1055T) than the 1050Ti paired with 1055T, to now falling short by just 11%. How incredible is that!
We take a look at a single gaming benchmark, CS:GO as that was the only game I had data on my with older processor. Again, the version of the game(post panorama update though), Operating System and GPU driver were much older for Phenom II X2 550 data - so not exactly apples to apples comparison.
The biggest gain was seen by from jumping to the Phenom hexa-core processor from its dual-core counterpart, at 160% increase in average frame rate and 143% in 1% low frame rate. The game went from being unplayable in Valve's official deathmatch (never mind the community ones) to significantly more playable when upgraded to Phenom II X6 1055T. Even so, CPU bottleneck was not entirely eliminated as we see further 42% increase in average frame rate and similar gain in 1% low frame rate too with 5600X at the helm.
Part 3: Test Setup and Methodology
Comments
Post a Comment