AMD FX-8800P vs Intel Core i5-3470T Comparison

The Intel Core i5-3470T and AMD FX-8800P make for an unusual pairing on paper: a 35 W desktop Ivy Bridge chip with two cores against a 15 W mobile Carrizo part with four cores, released roughly three years apart. Yet the recorded data shows them landing in nearly the same overall performance band, with average benchmark scores of 863 for the Intel part and 843 for the AMD part. The head-to-head record splits 6 to 1 in Intel's favor, but the one AMD win is sizable, and that asymmetry is where the interesting story lives.

Head-to-Head Benchmarks

The Cinebench results are remarkably uniform. In Cinebench R15 multi-core, the Core i5-3470T scores 256 against 242 for the FX-8800P, a 5.8 percent advantage. Cinebench R20 multi-core repeats the pattern exactly: 1069 versus 1010, again 5.8 percent. Cinebench R23 multi-core lands at 2547 versus 2407, a 5.8 percent gap for the third time in a row. That consistency across three generations of the same rendering test suggests the gap is structural, not noise. It shows up in the single-core numbers too: the Intel chip scores 150 versus 142 in R20 single-core (5.6 percent) and 359 versus 339 in R23 single-core (5.9 percent).

Geekbench single-core follows the same direction, with the i5-3470T at 560 against 524 for the FX-8800P, a 6.9 percent lead. This is the widest single-threaded margin in the dataset, and it aligns with what the clock figures imply: the Intel part boosts to 3.60 GHz while the AMD part tops out at 3.40 GHz, and the older Excavator core carries a lower instructions-per-clock reputation that these results bear out.

Then there is the outlier. Geekbench multi-core goes to the FX-8800P, 1234 to 1102, a 10.7 percent win, and it is the only result in the entire head-to-head set where the AMD chip finishes ahead. Why would a four-core processor lose every Cinebench multi-core test by nearly 6 percent yet win Geekbench multi-core by double digits? Geekbench consists of many short, varied workloads, and its scoring appears to reward the FX-8800P's four physical cores in bursts. Cinebench's sustained rendering load, by contrast, favors the Intel chip's stronger per-thread throughput, which its four threads (on two cores, with Hyper-Threading) evidently sustain well enough to stay ahead.

Where Each One Wins

For pure rendering workloads, the data is unambiguous. The Core i5-3470T wins every Cinebench test, multi-core and single-core alike, by margins between 5.6 and 5.9 percent. If the workload is rendering, video encoding of a similar character, or any application where per-thread speed dominates, the Intel part is the recorded leader, and its 6.9 percent Geekbench single-core edge reinforces that conclusion for lightly threaded general computing.

The FX-8800P's case rests on one number: the 10.7 percent Geekbench multi-core victory, the largest margin in either direction across the whole comparison. It is the only measured scenario where its four cores convert into a win, and it hints that short, parallel, bursty workloads could favor the AMD design. Notably, it achieves this within a 15 W TDP, less than half the 35 W envelope of the Intel chip, which matters for thermally constrained systems even though the comparison spans different sockets and market segments.

Architecture Differences

These chips could hardly be more architecturally distinct. The Core i5-3470T is an Ivy Bridge design built on Intel's 22 nm process, with a 93.6 mm² die, two cores, and four threads. The FX-8800P is an Excavator-based Carrizo design fabricated by GlobalFoundries on 28 nm, carrying 3,100 million transistors on a much larger 250 mm² die. The AMD part is newer by roughly three years, released in 2015 versus 2012 for the Intel chip, yet the newer process node is actually the coarser of the two, which is unusual and worth noting: the mobile Carrizo chip was built at 28 nm while the older desktop part enjoyed 22 nm.

The cache philosophies diverge sharply. Intel gives each core 64 KB of L1 and 256 KB of L2, backed by 3 MB of shared L3. The FX-8800P has 320 KB of L1 and 1 MB of L2 per module, with no L3 cache at all. That missing last-level cache is a plausible contributor to the sustained-rendering deficits, since Cinebench workloads tend to be cache-hungry.

Integrated graphics also differ meaningfully. The Intel part carries HD 2500 graphics; the AMD part integrates Radeon R7 graphics with 8 compute units, and AMD's integrated GPU configurations of that class were generally the stronger proposition, though the database records no GPU benchmarks here, so that advantage is qualitative only.

Specification Differences

• Segment and socket: The i5-3470T is a desktop chip on Intel Socket 1155; the FX-8800P is a mobile chip on AMD Socket FP4. They cannot share a platform.
• Cores and threads: 2 cores / 4 threads for Intel versus 4 cores / 4 threads for AMD.
• Clocks: 2.90 GHz base and 3.60 GHz boost for the Intel part, against 2.10 GHz base and 3.40 GHz boost for the AMD part.
• TDP: 35 W for the Intel chip versus 15 W for the AMD chip.
• Cache: 3 MB shared L3 on the Intel side; no L3 on the AMD side, with different L1 and L2 organization as detailed above.
• PCIe: Gen 3 with 16 CPU lanes on the Intel part versus Gen 3 with 8 CPU lanes on the AMD part.
• Memory bandwidth: Recorded at 34.1 GB/s for the FX-8800P; no figure is recorded for the i5-3470T. Both support dual-channel DDR3 and neither supports ECC.
• Production status: The FX-8800P is listed as end-of-life; no status is recorded for the i5-3470T.
• Release dates: 2012 for the Intel chip, 2015 for the AMD chip.

FAQ

Q: Which CPU is faster overall?

A: By the recorded averages, the Core i5-3470T leads, 863 to 843, and it wins 6 of 7 head-to-head benchmarks. However, the FX-8800P wins Geekbench multi-core by 10.7 percent, so the answer depends on the workload.

Q: Which one is better for rendering?

A: The Core i5-3470T. It wins Cinebench R15, R20, and R23 multi-core tests by 5.8 percent each time, plus both single-core Cinebench tests.

Q: Does the FX-8800P win anything?

A: Yes, Geekbench multi-core, 1234 to 1102, a 10.7 percent victory and the largest margin in the dataset.

Q: Which chip is more power efficient?

A: The FX-8800P has a 15 W TDP versus 35 W for the i5-3470T, and it delivers comparable average performance, which suggests strong efficiency for its envelope, though the database does not record measured power draw.

Q: Which CPU is in the higher percentile?

A: Nearly tied: the i5-3470T sits at the 23rd percentile against all CPUs, the FX-8800P at the 22nd.

Q: Do they compete with similar rivals?

A: Their rival clusters differ. The i5-3470T's nearest rivals include the Xeon W3520 and A8-7650K, while the FX-8800P's include the Core i7-3520M (an exact tie in average score) and the Core i7-920, placing it against mobile and older high-end parts.

The Verdict

The data points to the Core i5-3470T as the more consistent performer: six wins from seven tests, a higher average score, and uniform 5 to 6 percent leads across every Cinebench discipline plus a 6.9 percent Geekbench single-core advantage. For rendering and single-threaded work, it is the recorded choice.

The FX-8800P argues for itself on two grounds: a 10.7 percent Geekbench multi-core win that suggests real burst-parallel capability, and a 15 W TDP that delivers nearly equivalent average performance at well under half the Intel chip's power budget. Its Radeon R7 integrated graphics with 8 compute units also compare favorably on paper to HD 2500, though no GPU measurements back that up here. For thermally limited or battery-driven scenarios where short parallel bursts dominate, the AMD part has a data-backed case. For everyone else, the Intel chip's broader win record makes it the safer pick in this dataset.