AMD Radeon HD 8690M vs AMD Radeon R7 350 Comparison

FAQ

Q: Which GPU has the higher average benchmark score?

A: The AMD Radeon R7 350, with an average benchmark score of 7425, compared to the AMD Radeon HD 8690M's 6137. This places the R7 350 in the 40th percentile of all GPUs, while the HD 8690M sits in the 36th percentile.

Q: How do the two compare in the Geekbench OpenCL test?

A: The R7 350 scores 7792, which is 27% higher than the HD 8690M's 6137. This is the only benchmark where both GPUs have recorded scores.

Q: Does the Radeon HD 8690M win in any benchmark category?

A: No. In the head-to-head benchmarks, the R7 350 wins 1 test and the HD 8690M wins 0. The HD 8690M has no recorded Vulkan score, while the R7 350 has a Geekbench Vulkan score of 7,057.

Q: What are the closest rivals to each GPU according to the database?

A: The R7 350's nearest rival is the Intel UHD Graphics 750, with an average score of 7,441, which is 0.2% lower. The HD 8690M's nearest rival is the Intel Iris Pro Graphics 6200, with an average score of 6,117, which is 0.3% higher.

Q: What is the manufacturing process for both GPUs?

A: Both are built on TSMC's 28 nm process node. The R7 350 uses the Cape Verde chip, while the HD 8690M uses the Sun chip, and both are based on the GCN 1.0 architecture.

Q: When were these GPUs released?

A: The AMD Radeon HD 8690M was released on 2013-02-28, while the AMD Radeon R7 350 was released later on 2016-07-05. Both are now end-of-life products.

Architecture Differences

Both GPUs share the same fundamental GCN 1.0 architecture and are fabricated by TSMC on a 28 nm process, but they are built around very different chips. The R7 350 uses the Cape Verde chip, which contains 1,500 million transistors on a 123 mm² die, resulting in a transistor density of 12.2 million per mm². In contrast, the HD 8690M uses the much smaller Sun chip, with 690 million transistors on a 56 mm² die, giving a slightly higher transistor density of 12.3 million per mm².

The R7 350 is the more fully featured part in terms of compute resources. It houses 512 shading units, 32 texture mapping units, and 16 raster operation pipelines. The HD 8690M, by comparison, is equipped with 320 shading units, 20 TMUs, and only 8 ROPs. This means the R7 350 has 60% more shading units, 60% more TMUs, and double the ROP count of the HD 8690M, which directly explains its higher pixel and texture throughput.

The memory subsystem also differs significantly. The R7 350 has a 128-bit memory bus, while the HD 8690M is limited to a 64-bit bus. Both use 2 GB of GDDR5 memory, but the R7 350's memory runs at an effective 4.5 Gbps, delivering 72.00 GB/s of bandwidth, whereas the HD 8690M's memory operates at 4 Gbps effective, yielding only 32.00 GB/s. The R7 350 also has a higher memory clock of 1125 MHz, while the HD 8690M's memory clock is 1000 MHz.

The R7 350's clock speeds are not listed in the database for base or boost, but the HD 8690M has a base clock of 925 MHz and a boost clock of 975 MHz. Even without a direct clock comparison, the R7 350 reaches 819.2 GFLOPS of FP32 performance, far above the HD 8690M's 624.0 GFLOPS. The pixel rate tells the same story: 12.80 GPixel/s versus 7.800 GPixel/s, and texture rates of 25.60 GTexel/s versus 19.50 GTexel/s.

In terms of interface and physical design, the R7 350 is a single-slot card with no power connectors, a suggested PSU rating of 250 W, and a length of 168 mm (6.6 inches, one inch shorter than the unspecified length of the HD 8690M in the database). The R7 350 uses PCIe 3.0 x16, while the HD 8690M uses PCIe 8 lanes, and the R7 350 offers display outputs of 1x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2; the HD 8690M has no recorded display outputs listed.

Both support DirectX 12 (11_1), OpenGL 4.6, and Vulkan 1.2.170, so API compatibility is identical. The R7 350 belongs to the Pirate Islands (R7 300) generation, with the HD 8690M in the Solar System (HD 8600M) generation, and both have predecessors (Volcanic Islands for the R7 350 and London for the HD 8690M, with successors Arctic Islands and Gem System respectively) that are noted in the database.

The R7 350's TDP is 55 W, though the HD 8690M has no TDP listed. There is no launch MSRP for either product in the database.

Where Each One Wins

The data is clear: the R7 350 wins in every measured category where comparisons exist, and the HD 8690M wins nowhere. The HD 8690M has no wins at all in head-to-head benchmarks, and its only benchmark score is 6137 in Geekbench OpenCL, a test the R7 350 leads by 27%. The R7 350 also has a Geekbench Vulkan score of 7,057, a benchmark the HD 8690M has not recorded, which is why the R7 350 has a wider benchmark footprint in the database.

The R7 7800 series, represented by the R7 350, is positioned well above the HD 8690M in the overall GPU hierarchy. The R7 350's nearest rivals, Intel UHD Graphics 750 at 7,441 and the HD 8690M's closest, Intel Iris Pro Graphics 6200 at 6,117, show the gap between them is roughly 1,300 points in average score terms. Meanwhile, rivals like the NVIDIA GeForce GTX 1650 average 7,747, only 0.6% higher than the R7 350, and the AMD Radeon HD 6950 average 6,210 is 1.2% higher than the HD 8690M.

A user needing higher pixel fill rates, more texture throughput, or better FP32 compute should choose the R7 350 without hesitation. It doubles the ROP count, offers 72.00 GB/s of memory bandwidth versus 32.00 GB/s, and has a 128-bit bus versus 64-bit bus. The HD 8690M is a smaller chip with fewer shading units and less memory bandwidth, and it trails the R7 350 by 27% in the only test that directly pits them, making it the weaker pick for raw performance tasks.

In any workload requiring Vulkan support, the R7 350 again has an advantage because the HD 8690M has no Vulkan score listed, while the R7 350 has a Geekbench Vulkan result of 7,057. There is no scenario in the data where the HD 8690M wins a benchmark, so the verdict is one-sided in favor of the R7 350 in all use cases where performance is the deciding factor.

The HD 8690M is a mobile-class GPU, but its performance profile, with 320 shading units and 32.00 GB/s of bandwidth, is far below the R7 350's desktop part. The R7 350 also has more display outputs and its PCIe 3.0 x16 interface, and it is a 168 mm long card with a single-slot design. The HD 8690M has no dimensions or slot width listed, but its smaller chip and lower bandwidth do not translate into a competitive benchmark position.

Neither GPU has a price listed in the database, so no value judgment is possible. What the metrics show is that the R7 350 is the only one of the pair with wins, and it wins decisively in every way the database records, leaving the HD 8690M without a single benchmark victory to claim, a fact that shapes the recommendation entirely around the R7 350.

Specification Differences

Below are the fields where the two AMD GPUs differ, based on the database, differ:

• Chip: Cape Verde (R7 350) versus Sun (HD 8690M, which has 690 million transistors versus 1,500 million and a die size of 56 mm² compared to 123 mm²).
• Transistor count: 1,500 million (R7 350) and 690 million (HD 8690M).
• Die size: 123 mm² and 56 mm²; the R7 350 has a transistor density of 12.2M / mm² and 12.3M / mm², the latter, 12.3M / mm², and the former 12.2M / mm², but the density of 12.2M / mm² is the R7 350's value; the HD 8690M's 12.3M / mm² is higher.
• Base clock: the HD 8690M has 925 MHz (R7 350 has no base clock listed).
• Boost clock: HD 8690M has 975 MHz, R7 350 has no boost clock listed.
• Memory clock: R7 350 is 1125 MHz, HD 8690M is 1000 MHz, with effective rates of 4.5 Gbps for the R7 350 and 4 Gbps for the HD 8690M.
• Memory bus width: 128-bit (R7 350) vs 64-bit (HD 8690M).
• Memory bandwidth: 72.00 GB/s for R7 350, 32.00 GB/s for HD 8690M.
• Shading units: 512 (R7 350), 320 (HD 8690M).
• TMUs: 32 (R7 350), 20 (HD 8690M).
• ROPs: 16 R7 350, 8 HD 8690M.
• Pixel rate: 12.80 GPixel/s (R7 350) and 7.800 GPixel/s (HD 8690M).
• Texture rate: 25.60 GTexel/s (R7 350) versus 19.50 GTexel/s (HD 8690M.
• FP32: 819.2 GFLOPS R7 350, 624.0 GFLOPS HD 8690.
• TDP: 55 W R7 350 (HD 8690M has no TDP value listed).
• Slot width: single-slot R7 350 (HD 8690M has no slot width).
• Power connectors: None R7 350, no connectors on HD 8690M, and the R7 350's suggested PSU 250 W is not matched, HD 8690M has no suggested PSU; PCIe 7 350 has a PCIe 3.0 x16, HD 8690M has PCIe 3.0 x8.
• Display outputs: R7 350 has 1x DVI 1x HDMI 1.4a 1x DisplayPort 1.2, HD 8690M has no display outputs.
• Dimensions R7 350 length: 168 mm 6.6 inches, HD 8690M does not list length, height, width, none for R7 350 has no height.
• Generation: The R7 350 is in the Pirate Islands (R7 300) generation, HD 8690M in Solar System (HD 8600M).
• releaseDate: R7 350 2016-07-05, HD 8690M 2013-02-28.
• Predecessor: R7 350 predecessor Volcanic Islands, HD 8690M predecessor London.
• Successor: R7 350 Arctic Islands, HD 8690M Gem System.
• Benchmarks R7 350 has two benchmarks: geekbench_opencl 7792 and geekbench_vulkan 7057, HD 8690M has geekbench_opencl 6137 only.
• PercentileVsAllGpus R7 350: 40, HD 8690M: 36.
• avgBenchmark: R7 350 7425, HD 8690M 6137.

Head-to-Head Benchmarks

The database only records a single head-to-head test between these two, the Geekbench OpenCL score, and it belongs to the AMD Radeon R7 350. The R7 350 scores 7,792, against the HD 8690M at 6,137, and the delta is 27% in favor of the R7 350. That 27% lead is the biggest win in the comparison, and it is also the only one, since the HD 8690M has no wins in head-to-head.

In context, the R7 350's Geekbench OpenCL score of 27% over the HD 8690M mirrors the overall standings, where the R7 350's average benchmark score is 7425 versus 6137, a gap that is roughly 1,200 points higher. The HD 8690M nearest rival, Intel Iris Pro Graphics 6200, at 6117, is only 0.3% behind the HD 8690M, while the R7 350's closest rival, Intel UHD Graphics 750, at 7441, is 0.2% lower than the R7 350, so the R7 350 sits in a higher tier of performance by a wider margin.

The difference in shading units, 512 versus 320, explains much of the gap: the R7 350 has a 60% higher count, and its FP32 of 819.2 GFLOPS versus 624.0 GFLOPS is a 31% advantage, both consistent with the 27% benchmark delta. The R7 350's memory bandwidth, 72.00 GB/s versus 32.00 GB/s, is more than double, which is a severe advantage for bandwidth-sensitive workloads, and its pixel rate of 12.80 GPixel/s is 64% higher than the 7.800 GPixel/s.

The HD 8690M has no Vulkan score, so the R7 350's Geekbench Vulkan result of 7,057 stands alone; no direct comparison is possible. But in the OpenCL test, the only true head-to-head, the R7 350 wins by 27%, a decisive margin that aligns with its higher transistor count (1,500 million versus 690 million) and larger die area (123 mm² versus 56 mm²).

There is no benchmark where the HD 8690M beats the R7 350. The wins column reads 1 for the R7 350 and 0 for the HD 8690M. The largest win each way is the same as the only win: the R7 350's 27% OpenCL victory, while the HD 8690M has no wins to report. The data does not support any other outcome.

The Verdict

The AMD Radeon R7 350 is the clear choice for anyone who needs the better performing GPU. It wins the only head-to-head benchmark by 27%, with a Geekbench OpenCL score of 7792 versus 6137, and it has a higher average benchmark score of 7425 versus 6137. The R7 350 also has a Geekbench Vulkan score of 7057, while the HD 8690M has no Vulkan score recorded, so it is also the only one of the two with Vulkan data.

Architecturally, the R7 350 has double the ROPs, 60% more shading units, and 60% more TMUs than the HD 8690M. It has a 128-bit memory bus carrying 72.00 GB/s of bandwidth, compared to the HD 8690M's 64-bit bus with 32.00 GB/s. Its pixel rate is 12.80 GPixel/s, its texture rate is 25.60 GTexel/s, and its FP32 is 819.2 GFLOPS, all far ahead of the HD 8690M's 7.800 GPixel/s, 19.50 GTexel/s, and 624.0 GFLOPS.

The HD 8690M is not without a place: it is a mobile GPU, with a boost clock of 975 MHz and a base clock of 925 MHz, and its smaller Sun chip consumes fewer transistors, but the database shows no benchmark where it wins. Its nearest rival, the Intel Iris Pro Graphics 6200, is only 0.3% behind it, meaning the HD 8690M is competitive with integrated graphics, not with the R7 350.

For a desktop user looking for a PCIe 3.0 x16 card with DVI, HDMI 1.4a, and DisplayPort 1.2 outputs, the R7 350 is the superior part, with higher transistor density per mm² (12.2 versus 12.3, essentially equal) and a 55 W TDP. The HD 8690M is end-of-life, and it has no wins in the database, so the verdict is a straightforward one: the R7 350 is the GPU to pick for better performance, while the HD 8690M remains an older, lower-performing option with no recorded advantage.