NVIDIA Upgrades The Quadro M6000 With 24 GB VRAM – Replaces The Original Variant With Twice The Memory at Same Price

Hassan Mujtaba
Posted 7 months ago

NVIDIA has released an updated version of their Quadro M6000 workstation graphics card. The updated version arrives a whole year later after the initial variant launched. The latest version of the Quadro M6000 replaces the previous card at the same price and comes with twice the VRAM buffer which was much asked by Quadro users since higher VRAM is an essential requirement in the workstation market with developers demanding a beefier VRAM buffer size.

The Quadro M6000 launched with 12 GB VRAM, has now been updated with 24 GB VRAM.

NVIDIA Quadro M6000 Updated With Twice The VRAM Buffer, Same $5000 US Pricing

The NVIDIA Quadro M6000 is a professional card aimed at the workstation and visualization market but has technical similarities to the GeForce GTX Titan X. In terms of specifications, the card is essentially the same as the previous Quadro M6000 model and packs a full GM200 core based on NVIDIA’s second generation Maxwell architecture.

The GM200 core on the Quadro M6000 comes with 6 Graphics Processing Clusters, each of which consist of four SMM (Streaming Multiprocessors Units) which have 128 cores in each SMM block. The GM200-400-A1 chip which is the fully enabled version of the core has all 24 SMM Units enabled which result in 3072 CUDA Cores, 192 Texture Mapping Units and 96 Raster operation units. The card features 3 MB L2 cache and has six 64-bit memory controllers which result in a 384-bit memory interface. The core clocks are maintained at 988 MHz at default frequency.

Slides depicting the performance aspects of the Quadro M6000 workstation card.

The memory is the only spec that has been upgraded on the latest version of the Quadro M6000. The original had 12 GB GDDR5 VRAM but the latest variant adds twice the VRAM buffer, increasing the memory to 24 GB. The memory still comes clocked at the standard speed of 6.6 GHz. The card pumps out 317.4 GB/s bandwidth. The card features a 225W TDP. Power is fed through a single 8-Pin connector and display outputs include the reference Maxwell configuration of DVI, HDMI and three display ports.

Unlike previous Quadro cards such as the K6000 itself, the NVIDIA Quadro M6000 features only 1/32 the DP (Double Precision) performance due to lack of FP64 hardware available on the Maxwell core architecture as reported by NVIDIA. Hence, the M6000 features just 0.2 TFlops of FP64 compute performance but since it is aimed at workstations where single precision matters most, the card packs an impressive 7.0 TFLOPs of FP32 compute performance.


Quadro M6000 Pushes the boundaries for visual creators at Sony Pictures Imageworks.

NVIDIA had redesigned the Quadro cards and finally ditching the older cooler shroud design in favor of the NVIDIA’s NVTTM cooler design. The Quadro M6000 has the same cooler as the Titan X with the only difference being the color scheme and a few connector panel replacements for stereo and power inputs. The card comes with a neat Green and Black color scheme with a black colored heatsink and cooler shroud. It features two SLI connectors to allow 4-Way GPU use in the most professional workloads.

The revised model will be available at the same price of $5000 US and this is currently the fastest workstation graphics card from NVIDIA. The competing product from AMD is their FirePro W9100 which is based on the Hawaii GPU and features 16 GB GDDR5 VRAM. It is possible that since higher capacity VRAM chips are now available in good quantities, AMD can introduce either a 32 GB variant or a Fiji based part with higher FP32 compute.NVIDIA Maxwell Quadro Lineup

NVIDIA’s Entire Maxwell Based Quadro Lineup.

They already have a 32 GB variant for server market in the form of the S9170 which features the Hawaii GPU core. AMD recently launched their Radeon Pro Duo graphics card which combines the features of both Radeon and FirePro markets. This card packs an insane 16 TFLOPs of compute performance is primarily built for workstations, developers and gamers. The card packs two Fiji GPUs and hence comes with a 4 GB per chip design which could be combined to 8 GB HBM in some specific workloads to harness the complete GPU power. With Pascal, NVIDIA is expected to fully unleash flagship chips based on HBM2 architecture that will have up to 32 GB HBM2 VRAM although those parts are expected to launch later in 2016 but an announcement is totally expected at this year’s NVIDIA GTC conference.

Workstation Generation Comparisons:

GPU NameNVIDIA Quadro M6000NVIDIA Quadro M6000AMD FirePro W9100AMD FirePro W9100NVIDIA Quadro K6000
GPU CoreGM200GM200Hawaii XTHawaii XTGK110
SMM/Compute Units2424444415
Core Count30723072281628162880
Core Clock1140 MHz1140 MHz930 MHz930 MHz900 MHz
FP32 Compute7.00 TFLOPs7.00 TFLOPs5.24 TFlops5.24 TFlops5.20 TFlops
Memory Bus384-Bit384-Bit512-Bit512-Bit384-Bit
Memory Clock6.6 GHz6.6 GHz5.0 GHz5.0 GHz6.0 GHz
Memory Bandwidth317.4 GB/s317.4 GB/s320 GB/s320 GB/s288 GB/s
Pricing$4999 US$4999 US$4999 US$3999 USN/A

Share on Facebook Share on Twitter Share on Reddit