CPU Comparison

The AMD A8-7680 and AMD A10-7890K are two legacy desktop processors sharing the same FM2+ socket, but they represent different architectural eras and power targets. The A8-7680 is a 45W Excavator part, while the A10-7890K is a 95W Steamroller part with a higher clock speed and more L2 cache. Despite these differences, benchmark data shows they deliver nearly identical performance, making the choice between them more nuanced than raw specs suggest.

FAQ

Q: Which processor has a higher average benchmark score?

A: The AMD A8-7680 has an average benchmark score of 1036, while the AMD A10-7890K scores 1035. This is a difference of just 0.1%, placing them statistically in a dead heat.

Q: Are there any differences in multi-core performance?

A: In Cinebench R23 multi-core, the A8-7680 scores 3011 versus 3009 for the A10-7890K, a 0.1% lead for the A8. In Cinebench R15 multi-core, both chips score exactly 303.

Q: Which chip has a higher boost clock?

A: The AMD A10-7890K has a boost clock of 4.30 GHz, which is 0.5 GHz higher than the A8-7680’s 3.80 GHz boost. The A10 also has a higher base clock at 4.00 GHz versus 3.50 GHz.

Q: How do the two chips compare in single-core performance?

A: They are effectively identical. In Cinebench R20 single-core, both score 178. In Cinebench R23 single-core, the A8-7680 edges ahead 425 to 424, a 0.2% difference.

Q: What is the power consumption difference?

A: The A8-7680 has a TDP of 45W, while the A10-7890K has a TDP of 95W. This makes the A8 substantially more power-efficient on paper.

Q: Which chip is unlocked for overclocking?

A: The AMD A10-7890K has an unlocked multiplier, while the A8-7680 does not. This gives the A10 a potential manual tuning advantage, despite its higher stock power draw.

Architecture Differences

The two processors come from different architectural families, despite sharing the same 28 nm process node from GlobalFoundries. The A8-7680 is based on the Excavator architecture with the Carrizo codename, while the A10-7890K uses the Steamroller architecture with the Godaveri codename. This is a significant generational split, as Excavator is a refinement of the earlier Steamroller design.

The transistor count reflects this architectural gap. The A8-7680 packs 3,100 million transistors on a 250 mm² die, while the A10-7890K has 2,411 million transistors on a slightly smaller 245 mm² die. The newer Excavator design manages a higher transistor density, which partially explains why the A8-7680 can match the A10’s performance at a much lower TDP.

Cache configurations also differ notably. The A8-7680 has a 320 KB L1 cache and 2 MB of L2 cache, while the A10-7890K has a smaller 256 KB L1 but doubles the L2 to 4 MB. Neither chip has any L3 cache, and both rely on DDR3 memory with dual-channel support and an identical 34.1 GB/s memory bandwidth.

Both chips feature integrated Radeon R7 graphics and support PCIe Gen 3, though the A10-7890K’s PCIe listing specifies 16 lanes for the CPU only. The A8-7680 does not have this lane specification in its data. The A10-7890K is marked as end-of-life production status, while the A8-7680 remains active in the product stack.

Head-to-Head Benchmarks

The benchmark results show an extraordinary level of parity between these two chips, with the A8-7680 winning all five head-to-head comparisons by razor-thin margins. The largest delta is just 0.2%, making this one of the closest matchups in the database.

In Cinebench R15 multi-core, both processors score exactly 303, with the A8-7680 named the winner by a 0% delta. The A10-7890K cannot separate itself even in a test that typically favors higher clock speeds. Moving to Cinebench R20 multi-core, the A8-7680 scores 1264 against 1263 for the A10, a 0.1% margin that is within run-to-run noise.

Single-core results follow the same pattern. Cinebench R20 single-core shows a perfect tie at 178 points for both chips. Cinebench R23 single-core gives the A8-7680 a 425 to 424 victory, another 0.2% difference. The A8-7680 also leads in Cinebench R23 multi-core by 3011 to 3009, again a 0.1% gap.

Given the A10-7890K’s substantial clock advantage, 0.5 GHz higher boost and 0.5 GHz higher base, one might expect it to win at least one test. The data shows the opposite: the A8-7680’s newer Excavator architecture compensates entirely for the clock deficit. In fact, the A8-7680 wins all five head-to-head tests, while the A10-7890K records zero wins.

Looking at the broader competitive landscape, both chips sit in the 28th percentile of all CPUs. The A8-7680’s nearest rivals include the A10-7890K itself (0.1% ahead), the Intel Core i7-4558U (-0.2% behind), the AMD A10-9700 (0.2% ahead), and the AMD A12-9800E (0.3% ahead). The A10-7890K’s rival list is nearly identical, with the A8-7680 appearing as its closest competitor at -0.1%.

Specification Differences

The core and thread counts are identical at 4 cores and 4 threads, so the differences lie elsewhere. The most obvious gap is clock speed: the A10-7890K runs at 4.00 GHz base and 4.30 GHz boost, versus 3.50 GHz base and 3.80 GHz boost for the A8-7680. That is a 0.5 GHz advantage for the A10 in both modes.

Power consumption is the reverse story. The A8-7680 draws only 45W TDP, while the A10-7890K is rated at 95W TDP. This 50W difference makes the A8 far more suitable for compact or low-power builds, though the A10’s unlocked multiplier offers manual overclocking headroom that the A8 lacks.

Cache hierarchy differs as well. The A8-7680 has 320 KB of L1 and 2 MB of L2. The A10-7890K has 256 KB of L1 and 4 MB of L2. This means the A10 has double the L2 capacity, which is notable given how close the benchmark scores are.

The architectures are distinct: Excavator (Carrizo) for the A8-7680 versus Steamroller (Godaveri) for the A10-7890K. Transistor counts differ too, with the A8 at 3,100 million and the A10 at 2,411 million. Die sizes are close, at 250 mm² for the A8 and 245 mm² for the A10.

Production status also separates them. The A8-7680 is listed as Active, while the A10-7890K is End-of-life. The release dates are far apart: the A8-7680 launched on 2018-10-25, while the A10-7890K came out on 2016-01-10. Both support DDR3 memory, lack ECC support, and share the same 34.1 GB/s memory bandwidth.

The Verdict

The benchmark data makes one thing clear: for stock performance, there is no meaningful winner between these two chips. The A8-7680 leads in all five head-to-head tests, but the margins are 0.2% or less. In practical terms, a user would not notice any difference in Cinebench workloads between the A8-7680 and the A10-7890K.

The deciding factors are therefore not raw performance but secondary characteristics. The A8-7680 offers a 45W TDP versus 95W for the A10-7890K, making it the clear choice for power-sensitive systems or those with limited cooling. It is also the newer part, released in October 2018 versus January 2016, and remains in active production while the A10-7890K is end-of-life.

The A10-7890K’s only advantages are its higher clock speeds and unlocked multiplier. For an enthusiast willing to manually overclock, the A10 could potentially pull ahead of the A8 in sustained workloads, but the stock benchmarks show no such benefit. The A10’s larger 4 MB L2 cache also does not translate into any measurable win in the tested Cinebench scenarios.

For a new build or a system where power efficiency matters, the AMD A8-7680 is the logical pick. It matches the A10-7890K’s performance at less than half the TDP, uses a newer architecture, and is still an active product. For a user with an existing FM2+ board who specifically wants overclocking headroom, the A10-7890K remains viable, but the data shows its stock performance is indistinguishable from the A8-7680.

Both chips sit at the 28th percentile of all CPUs, meaning they are firmly entry-level parts by modern standards. Their average benchmark scores of 1036 and 1035 place them in a cluster that includes the Intel Core i7-4558U and the AMD A10-9700, all within 0.3% of each other. Neither chip offers a compelling reason to choose it over the other based on performance alone, so the decision should rest on TDP, production status, and overclocking preference.