Skip to content

Can DGX Spark 64G run Ripple-65B-Preview?

Comfortable: GGUF Q5_K_M needs 50.8 GB at 4K, batch 1; headroom 13.2 GB.

Text input, FP16 KV cache, one card. Unified-memory capacity uses the GPU guide’s usable limit; negative headroom may require host offload. Speed source settings may differ from this 4K estimate.

Measured runs

No measured run for this pair. Sizing is an estimate, not a runtime test.

Quant memory at 4K

QuantTotal GBVerdictWeight sizing
GGUF IQ1_M20.0ComfortableVerified file
GGUF IQ2_XXS21.6ComfortableVerified file
GGUF IQ2_XS23.8ComfortableVerified file
GGUF IQ2_S27.2ComfortableVerified file
GGUF IQ3_XXS29.9ComfortableVerified file
GGUF IQ3_S34.5ComfortableVerified file
GGUF IQ4_XS41.1ComfortableVerified file
GGUF Q4_K_M41.2ComfortableVerified file
GGUF Q5_K_M50.8ComfortableVerified file
GGUF Q6_K60.9TightVerified file
GGUF Q8_072.6Does not fitVerified file

Context capacity · GGUF Q5_K_M

ContextTotal GBVerdict
4K50.8Comfortable
16K51.6Comfortable
32K52.7Comfortable
128K59.3Tight

Generation speed

GGUF Q5_K_M: 6 tok/s · Bandwidth ceiling (not measured)

Other model / GPU pairs