Accelerator
NVIDIA GeForce 8800 GTX: 768 MiB of GDDR3, 86.4 GB/s peak bandwidth, and 518 GFLOP/s FP32.
The G80 replaced separate pixel and vertex units with a single pool of general-purpose shaders. That is what made the card programmable enough for general computation, and CUDA followed within a year.
| Component | Role | Count | Clock | Capacity |
|---|---|---|---|---|
| NVIDIA Tesla (NVIDIA GeForce 8800 GTX) Reference-card graphics clock. | GPU | 1 | 575 MHz Nominal clock | — |
| NVIDIA GeForce 8800 GTX frame buffer | Video memory | 1 | — | 768 MiB |
provisional A figure marked provisional is recorded but not accepted: its evidence is unsettled. It may rely on one source, an unconfirmed announcement, or a digit read from a damaged scan. The reason appears under the figure. The catalog shows the figure but never uses it in a multiplier. How records are judged.
| Metric | Value | Evidence | Basis | Method | Source |
|---|---|---|---|---|---|
| Memory bandwidth whole-system provisional | 86.4 GB/s | Confirmed | Derived | Theoretical peak Computed here from figures the vendor stated Adds the one fitted frame-buffer bandwidth. No host memory is included. |
provisional A figure marked provisional is recorded but not accepted: its evidence is unsettled. It may rely on one source, an unconfirmed announcement, or a digit read from a damaged scan. The reason appears under the figure. The catalog shows the figure but never uses it in a multiplier. How records are judged.
| Metric | Value | Evidence | Basis | Method | Source |
|---|---|---|---|---|---|
| Clock frequency gpu provisional | 575 MHz | Reported | Vendor-rated | Nominal clock Stated by the vendor A stock GeForce 8800 GTX in the independent technical comparison, not a factory-overclocked partner board. | |
| Peak FP32 rate gpu provisional | 518 GFLOP/s | Confirmed | Theoretical | Theoretical peak Stated by the vendor General shader units at the vendor-stated reference clock. A theoretical general-shader FP32 peak; tensor, AI and ray-tracing rates are excluded. |
| Metric | Value | Evidence | Basis | Method | Source |
|---|---|---|---|---|---|
| Thermal design power gpu provisional | unknown Neither qualifying source supplies a numeric GeForce 8800 GTX board-power or thermal-design rating. | Not stated | Vendor-rated | Rated power Stated by the vendor Neither qualifying source states a numeric board-power or thermal-design rating. |
provisional A figure marked provisional is recorded but not accepted: its evidence is unsettled. It may rely on one source, an unconfirmed announcement, or a digit read from a damaged scan. The reason appears under the figure. The catalog shows the figure but never uses it in a multiplier. How records are judged.
| Metric | Value | Evidence | Basis | Method | Source |
|---|---|---|---|---|---|
| Memory bandwidth memory provisional | 86.4 GB/s | Confirmed | Theoretical | Theoretical peak Stated by the vendor Vendor-stated maximum transfer rate of this memory interface, not a benchmarked application result. | |
| Memory capacity whole-system provisional | 768 MiB | Reported | Vendor-rated | Design capacity Stated by the vendor The frame buffer fitted to the reference accelerator; host memory is excluded. |
86 400 000 000 B/s
a + b + … sum v1 Adds 1 memory-bandwidth figure obtained by theoretical-peak. Nothing in the catalog is left out of it. A total is only as trustworthy as the decision about what belongs in it, so both the terms and the exclusions are named records.
This accelerator’s whole-system memory is its fitted frame buffer; host memory is excluded. Partner cards may differ from this reference configuration.