AMD Radeon HD 8870M vs AMD Radeon R7 350 Comparison

FAQ

Q: Which GPU is faster in OpenCL compute workloads?

A: The AMD Radeon HD 8870M wins the only shared benchmark, Geekbench OpenCL, with a score of 8462 versus 7792 for the Radeon R7 350. That is a delta of 8.6% in favor of the HD 8870M.

Q: How do the two GPUs compare in overall benchmark standing?

A: The HD 8870M sits at the 43rd percentile among all GPUs in the database, while the R7 350 sits at the 40th percentile. The HD 8870M's average benchmark score is 8462, and the R7 350's average is 7425.

Q: What is the closest rival to each card according to the database?

A: The HD 8870M's nearest rival is the NVIDIA GeForce MX330, which scores 8458, a delta of just 0.1%. The R7 350's nearest rival is the Intel UHD Graphics 750, which scores 7441, a delta of -0.2%.

Q: Do both cards use the same memory configuration?

A: Yes. Both come with 2 GB of GDDR5 memory on a 128-bit bus, with 1125 MHz memory clock (4.5 Gbps effective) and 72.00 GB/s bandwidth. Memory is identical between the two.

Q: Which card has a higher shading unit count?

A: The HD 8870M has 640 shading units, while the R7 350 has 512 shading units. The HD 8870M also has 40 texture mapping units versus 32 on the R7 350, but both have 16 ROPs.

Q: Are both cards from the same architecture generation?

A: Both use GCN 1.0 architecture and are built on TSMC's 28 nm process with 1,500 million transistors on a 123 mm² die. However, they belong to different product generations: the HD 8870M is part of the Solar System (HD 8800M) generation, while the R7 350 is part of the Pirate Islands (R7 300) generation.

Architecture Differences

Both GPUs are built on GCN 1.0 architecture, which means they share core design philosophies from AMD's first-generation Graphics Core Next. They are both fabricated by TSMC on a 28 nm process, with identical transistor counts of 1,500 million and identical die sizes of 123 mm². The transistor density works out to 12.2M per mm² on both chips. The underlying chip names differ: the HD 8870M uses the Venus chip, while the R7 350 uses Cape Verde. Despite the different branding and generations, the silicon heritage is closely related.

The two cards belong to different market generations. The HD 8870M is from the Solar System generation, specifically the HD 8800M family, and was released on March 31, 2013. Its predecessor is London and its successor is Gem System. The R7 350, by contrast, comes from the Pirate Islands generation, the R7 300 family, with a release date of July 5, 2016. Its predecessor is Volcanic Islands and its successor is Arctic Islands. The production status for both is end-of-life.

The compute resources differ notably. The HD 8870M packs 640 shading units, 40 TMUs, and 16 ROPs. The R7 350 has 512 shading units, 32 TMUs, and 16 ROPs. The R7 350 therefore has 20% fewer shading units and 20% fewer texture units, while maintaining the same raster output stage count. This configuration difference drives most of the performance gap.

Clock speeds are only partially documented for the R7 350. The HD 8870M has a base clock of 725 MHz and a boost clock of 775 MHz. The R7 350's base and boost clocks are not recorded in the database. Memory clocks are identical at 1125 MHz, translating to 4.5 Gbps effective on both cards.

The R7 350 has a defined power envelope with a TDP of 55 W, a suggested PSU of 250 W, and no power connectors, indicating a slot-powered design. It is also listed as a single-slot card with a length of 168 mm (6.6 inches). The HD 8870M does not have TDP, slot width, or power connector data in the database, likely because it is a mobile part.

The R7 350 includes display outputs of 1x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2. The HD 8870M's display outputs are not recorded. Both cards use a PCIe 3.0 x16 bus interface.

API support is identical: DirectX 12 (11_1), OpenGL 4.6, and Vulkan 1.2.170. Neither card has ray tracing cores or tensor cores. Neither has a recorded FP16 performance figure. FP32 output favors the HD 8870M at 992.0 GFLOPS versus 819.2 GFLOPS for the R7 350.

Where Each One Wins

The HD 8870M is the clear winner in raw compute throughput. Its FP32 figure of 992.0 GFLOPS is 21% higher than the R7 350's 819.2 GFLOPS. Its texture rate of 31.00 GTexel/s outpaces the R7 350's 25.60 GTexel/s by a similar margin. The pixel rate is the one rasterization stat where the R7 350 edges ahead: 12.80 GPixel/s versus 12.40 GPixel/s. That small lead, 3.2%, comes from the R7 350's higher clock behavior, though the base and boost clocks are not recorded for that card.

In the only head-to-head benchmark recorded, Geekbench OpenCL, the HD 8870M wins with 8462 against 7792, an 8.6% advantage. The HD 8870M also holds a higher percentile ranking, 43rd versus 40th, and a higher average benchmark score of 8462 versus 7425.

The R7 350, however, has an advantage in the Vulkan API. It has a recorded Geekbench Vulkan score of 7057, while the HD 8870M has no Vulkan benchmark entry in the database. That means the R7 350 is the only one of the two with verified Vulkan compute performance data.

The R7 350 also has a more complete physical specification set. It is documented as a 55 W card with a 250 W suggested PSU, no power connectors, single-slot thickness, and a 168 mm length. The HD 8870M lacks that physical data, which likely reflects its mobile-oriented design. For a desktop build where power draw and slot dimensions matter, the R7 350 offers concrete numbers.

For users seeking the strongest OpenCL compute performance, the HD 8870M is the pick. For users who need Vulkan benchmark data or a fully documented desktop card with a known power envelope, the R7 350 is the more complete package. The R7 350 also has a longer production lifespan in terms of release date, launching in 2016 versus 2013 for the HD 8870M.

Specification Differences

The two cards differ in several key specification fields. The HD 8870M uses the Venus chip, while the R7 350 uses Cape Verde. Both are GCN 1.0, but the HD 8870M belongs to the Solar System (HD 8800M) generation, whereas the R7 350 belongs to the Pirate Islands (R7 300) generation. The HD 8870M was released on March 31, 2013; the R7 350 was released on July 5, 2016.

Shading units differ: 640 on the HD 8870M versus 512 on the R7 350. TMUs differ: 40 versus 32. ROPs are the same at 16. The HD 8870M has a base clock of 725 MHz and a boost clock of 775 MHz; the R7 350 has no recorded base or boost clock. Memory clocks are the same at 1125 MHz (4.5 Gbps effective).

Pixel rate slightly favors the R7 350 at 12.80 GPixel/s versus 12.40 GPixel/s. Texture rate favors the HD 8870M at 31.00 GTexel/s versus 25.60 GTexel/s. FP32 compute favors the HD 8870M at 992.0 GFLOPS versus 819.2 GFLOPS.

The R7 350 has a TDP of 55 W, a suggested PSU of 250 W, and no power connectors. The HD 8870M has no TDP, power connector, or PSU data. The R7 350 is single-slot and 168 mm long (6.6 inches). The HD 8870M has no dimension data. The R7 350 has display outputs of 1x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2. The HD 8870M has no display output data.

The R7 350 has two benchmark entries: Geekbench OpenCL at 7792 and Geekbench Vulkan at 7057. The HD 8870M has one: Geekbench OpenCL at 8462. The average benchmark score is 8462 for the HD 8870M and 7425 for the R7 350. Percentile rankings are 43 versus 40.

The R7 350's predecessor is Volcanic Islands and its successor is Arctic Islands. The HD 8870M's predecessor is London and its successor is Gem System. Neither card has a launch MSRP recorded in the database. Memory size, type, bus width, and bandwidth are identical: 2 GB GDDR5, 128-bit, 72.00 GB/s. Transistor count, die size, foundry, and process node are identical as well.

Head-to-Head Benchmarks

The only direct comparison in the database is Geekbench OpenCL. The HD 8870M scores 8462, and the R7 350 scores 7792. The HD 8870M wins by 8.6%. Contextualizing that number against the nearest rivals helps clarify what it means. The HD 8870M's nearest rival, the NVIDIA GeForce MX330, scores 8458, a delta of 0.1% ahead of the HD 8870M. The Intel Arc A380 scores 8558, which is 1.1% above the HD 8870M. The AMD Radeon 880M scores 8436, 0.3% below. The NVIDIA GeForce GTX 675MX scores 8427, 0.4% below. The HD 8870M is clustered tightly among these four rivals, with all deltas within roughly 1 percentage point.

The R7 350's OpenCL result of 7792 places it near its own rival cluster. The Intel UHD Graphics 750 scores 7441, which is 0.2% below the R7 350. The AMD Radeon HD 8850M scores 7447, 0.3% below. The NVIDIA GeForce GTX 1650 scores 7472, 0.6% below. The Intel Arc A310 scores 7550, 1.7% below. The R7 350 leads all four of its nearest rivals, but by small margins. The gap between the two cards in question, 8.6%, is far larger than the gaps between each card and its own closest competitors. That suggests the HD 8870M sits in a meaningfully higher performance tier in OpenCL compute.

The R7 350 has an additional benchmark, Geekbench Vulkan, where it scores 7057. There is no Vulkan score for the HD 8870M, so no direct comparison is possible. The R7 350's Vulkan result is lower than its OpenCL result by roughly 9.4%, which indicates that its Vulkan compute performance trails its OpenCL performance on the same card.

The average benchmark score paints a broader picture. The HD 8870M's average is 8462, while the R7 350's average is 7425. That is a difference of 1037 points, or about 14% lower for the R7 350. The average score for the R7 350 is dragged down by its Vulkan result, which is below its OpenCL result.

In terms of percentile standing, the HD 8870M ranks at the 43rd percentile across all GPUs, while the R7 350 ranks at the 40th percentile. A three-percentile gap is modest, but it aligns with the benchmark deltas.

The head-to-head data shows a clear pattern: the HD 8870M is the faster card in OpenCL compute, with a double-digit GFLOPS advantage, a higher texture rate, and a higher average benchmark score. The R7 350 counters with a slight pixel rate lead, a Vulkan benchmark entry, and a fully documented desktop power profile. For compute workloads measured by Geekbench OpenCL, the HD 8870M leads by 8.6%. For users focused on Vulkan compute, the R7 350 at least has recorded data, even if no direct comparison exists. The overall conclusion from the database is that the HD 8870M holds the performance edge, while the R7 350 offers a more complete and self-contained desktop package.