• expired

[Prime] Corsair Vengeance RGB DDR5 RAM 96GB (2x48GB) 6000MHz CL36 $311 Delivered @ Amazon AU

630
This post contains affiliate links. OzBargain might earn commissions when you click through and make purchases. Please see this page for more information.

Original Coupon Deal The extra 5% PAYTOPD coupon from here has been fully redeemed unfortunately. Still a good deal at $311.

Low price for 96 GB (2x48 GB) of RGB DDR5 RAM at 6000 MHz CL36, which of course makes it better, because RGB than the previous deal for 96 GB of the plain old no RGB Corsair RAM 6000 Mhz CL36.

Update: On backorder now.

Out of Stock More importantly, the $295 Prime Deal for 96 GB of non-RGB DDR5 RAM is back in stock, so get it while you can during the Prime Day Sales I guess.

Price History at C CamelCamelCamel.
This is part of Amazon Prime Day sales for 2025

Related Stores

Amazon AU
Amazon AU
Marketplace

Comments

  • +19

    Why did the OzBargainer buy 96GB of Corsair RGB RAM?
    Because it was cheaper than therapy and lights up when you cry.

    • +10

      Jokes on you, I do all my crying in the bathroom.

      • +1

        Bathroom PC?

        • +6

          No, bathroom only PP

        • +2

          you guys have bathroom?
          sorry, need to get out of bathroom as boss is shouting…

        • +1

          Use case requires solid state, not random access

    • +1

      I'm not planning to buy but tested as well and can confirm I get the same error The promotional code you entered is not valid.

    • Huh, it was showing up for me previously ( like 40 minutes ago), but I think the original coupon deal with PayTo Payment is over? Thats unfortunate.

  • +1

    RGB version is definitely faster.

    • +1

      Don't tell our secret.

    • +2

      Specifically the red

  • +4

    probably samsung or maybe even micron if you're super unlucky, might want to pay a bit extra to guarantee hynix, these samsung sticks are already at their limit, but i've gotten hynix 6000/30 2x48 sticks to run at 8000/cl38 on Z890, so hynix sticks are definitely more future proof (i.e if you upgrade to zen6 somewhere down the line hynix will overclock 8000+ easily, currently they're just choked by CPU's IMCs, meanwhile these won't even reach 6400 on a good day)

  • +4

    Should be enough to run Chrome on a Windows PC

  • +1

    Any issues mixing RAM?

    I have 2x32GB DDR5 RAM at 6,000 MHz CL32.
    Im not too concerned about overclocking RAM, with diminishing returns on a B650 Mobo with the earlier Prime day Ryzen 5 7600x CPU

    • I'd probably just replace it and re-use/sell your existing?

      Different capacities, different timings. I wouldn't personally. Ryzen is funny about 4 sticks too. Likely have issues with expo/getting timings right etc

    • +3

      AM5 doesn't like mixing ram its barely stable with 2x 32GB when running EXPO / XMP :)

      Zen6 is rumored to have a dual memory controller to help manage larger sizes above 64GB :)

      • +3

        This is why I went for 2x 32 in my new build, parts started arriving today so looking forward to putting it all together.
        Kinda wish I'd ordered a 96GB kit now though… the model I bought appears to be hard to find now because the manufacturer has released a slightly revised version of it and I'm a bit worried that if I mix the two kits there will be stability issues.

        • +3

          The larger size the more stress on the memory controller. you may find your system more stable with 64 or even less ram :) not every CPU can do 64GB 6000 CL30

          The fix on those CPU's is do reduce speed to like 5600MT's or even keep it at stock 4800MT/s

          • @vid_ghost: I'm getting an MSI MPG X870E EDGE TI WIFI motherboard and a Ryzen 9 9900X so I hope it can handle 64GB just fine with EXPO enabled. The RAM is Silicon Power DDR5 6000 CL30.

            • +3

              @bonezAU: Its 100% CPU lottery.. I had a Ryzen 7500F run 2x32GB sticks @t 6000 CL30 just fine… change to 7700 and it wouldn't boot once a week and had to be reset or need to memory train once a week.

              :) Its a gamble.. Motherboard doesn't matter… some extra copper layers 6 or 8 layers can help a little but its mostly the CPU

              Pretty much all of the CPU's can do 32GB 6000 CL30 :) thats the most stable size and speed

              Lots of workstations with 128GB's of DDR5 or more are running at 4800MT/s CL 40

              • @vid_ghost: Thanks for the info. I'll experiment and see how far I can push it. Got most of the kit today, just waiting on the CPU cooler now. It's all sitting here staring at me waiting to be built!

          • @vid_ghost: any tips here? I picked up 2x32 CL30 expo ram, but memtest explodes when I try to load expo settings - stable at 4800, but I'm not sure how to come up with steps in between to test - and pretty annoyed that expo settings I paid for are apparently optional when the mobo claims it is capable!

            • +1

              @myr: It's got nothing to do with the RAM or your motherboard - they both do support it 100%.

              The problem is, the memory controller is on the CPU for Ryzen, and it's notoriously poor. It's only specced for 5200MT/s max for 7000 series and 5600MT/s for 9000 series, and anything over that is considered an overclock and isn't 'supported'. EXPO being 6000MT/s CL30 is pushing the MC hard and not all CPUs will cope, particularly 7000 series. You also can't RMA a Ryzen CPU for a poor memory controller, being you're running out of spec. It's just silicon lottery.

              People often don't talk about it, and I don't know why as it's Ryzen's biggest downside, although AGESA updates have improved memory stability heavily the last few years.

              My previous 7800X3D would only take 1 kit of RAM at EXPO out of 5 I tried. They all worked at base spec.

        • I ordered the non RGB 96GB kit and will be testing it in my ITX build with the ASUS ROG STRIX B650E-I motherboard. It’s most likely SK Hynix M-die, though there's a small chance it could be A-die, which would be a nice bonus. I’ll share an update if I run into any issues.

    • +1

      If you do 64+96 you’d be lucky to get 5200 mhz (same with 32x4 though)

    • +2

      AMD CPUs are barely stable with four matched sticks over 32GB, let alone mixed.

      • You said it brother. Last year end I had such horrendous time trying to get 64GB GSkill kit working on AMD 9800X3D on MSI X870E Carbon, that eventually after two months, I had to send the mobo back. Eventually I was able to attain stability with Gigabyte X870E Aorus Master and a different 64GB RAM Kit from ADATA. My entitre Christmas time went away trouble shoting the newly built PC rather then playing games arounf that time.

        • This is why I went back to Intel. Yeah the socket changing every generation sucks but at least my RAM and system works

  • -2

    Will it work with 7800x3d and b650m gaming wifi plus MB without crashing issues?

    • What crashing issue?

      • as I read so many posts about mentioning the AM5 could only handling less than 64G RAM

        • +1

          Correct, I had horrendous time trying to boot properly on my new AM5 platform for close to two and a half months roughly before a new combo of mobo and memory kit allowed me to boot without further issues. For a bit more info, see my above post.

    • +1

      'Can you promise me that if I pair this RAM with my machine it won't crash?'

      Is that what you're asking? lol. The answer is - we don't know. Too many factors can cause 'crashing' - my brothers computer is, far as I can tell, completely fine at a hardware level. Crashed on/off on IDLE ONLY (not under load) randomly 1-2 times a week. I just reinstalled /uinstalled GPU/chipset all the drivers a few times, using DDU as well, and appears to have stopped.. Sooooo we couldn't say there's no 'crashing issues' with this RAM for you.

      • thanks for the reply. but there are so many posts about mentioning the AM5 CPUs can only handling less than 64G RAM

    • I mean, there's no reason why there would be crashing issues with this ram. But if you're already getting those, it could be any other component causing it and swapping ram won't necessarily fix it.

    • Have no idea what this guys are going on about. I'm running a 7950x3d, with 96GB DDR5 6000, on an Asus x670e motherboard and have no issues. I also work in an industry where they are using these CPUs with 128gb memory without any issues.

  • -1

    which one better?

    with or without rgb

    • The RGB goes faster cause it looks pretty

    • As someone who currently has an RGB kit from GSkill DDR4….

      Unless you specifically want a colour theme going/everything is visible via side panel or something, just get without. It's one less thing to fail, and it's one less piece of software to manage as well potentially. I'm all in on the all-black/no lights look but I have an Ncase M1 ITX case. I have the glass side panel but for temps I stopped using it.

    • +1

      The RGB sticks can block a large CPU cooler from a proper fitting. There’s often not much in it and the little bit of extra height on the RGB sticks can be a problem.

  • Dang the CL30 version is still $200 more.

    • Darn, I just bought CL30 version of this yesterday for $529. Should I go with this instead?

      • you pay a premium for tighter timings mate.

        • I cancelled and bought this one instead. Perplexity research told me this was the better value, lol. This is what it said:

          Recommendation for Your Build
          Go with CL36 for Better Value
          Reasons to choose DDR5-6000 CL36:

          • Significant cost savings: $218 AUD less expensive
          • Negligible performance difference: 1-2% at most in real-world usage
          • Excellent compatibility: Works reliably with your motherboard/CPU
          • Future-proof capacity: 96GB handles large LLM models easily

          When to Consider CL30
          Only choose CL30 if:
          * Budget is not a concern and you want absolute maximum performance
          * You're building a showcase system where every specification matters
          * Price difference is minimal (under $50 AUD)

          Final Verdict
          For your RTX 5090 LLM inference workstation, DDR5-6000 CL36 is the smart choice. The $218 AUD savings can be better invested in other components, and the performance difference is practically unnoticeable in your intended use case.

          Recommended Kit: Corsair VENGEANCE RGB DDR5-6000 CL36 96GB (2×48GB) at $311 AUD from Amazon Australia

          This provides excellent value while maintaining full compatibility with your ASRock X870E Taichi Lite motherboard and AMD Ryzen 9 9950X3D processor for professional LLM inference workloads.

          • @ggbhai: if you're on a budget sure, the cost saving is massive. i already splurged on a 5090 suprim soc, i just went with the cl30 one, i use mine for gaming + stable diffusion mostly.

            • @plentifoo: Nice. Yes, I splurged as well and as a result getting poorer. I need to budget for another 5090 and 2x48 gb ram in this build in near future without getting divorced, lol. What config did you get? Mine was R9 9950X3D, ASROCK X870E Taichi Lite, Lian Li o11 dynamic evo XL, 2 TB Samsung PRO nvme primary and Cooler Master X Mighty Platinum 2000 2000w and Lian Li AIO cooler and SL infinity fans. I don't game, my primary usecase is AI development.

          • @ggbhai: This is CL36-44-44-96 but most cl36 is 36-36-36, is it still different by 2%?

            • @unknowname: This is what Perplexity research says:

              RAM Timing Performance Comparison: CL36-44-44-96 vs CL36-36-36-96 vs CL30-36-36-76

              1. CAS Latency (CL) and Real-World Impact

              CAS Latency (CL) is the primary figure affecting memory response time, but secondary timings (tRCD, tRP, tRAS) also play a role in total access latency and memory bandwidth. Lower numbers are generally better, but the difference is often small at the same frequency.

              2. Latency Calculations (DDR5-6000)
              Timing Set CAS Latency (ns) Total Primary Timings (sum)
              CL36-44-44-96 12.0 ns 220
              CL36-36-36-96 12.0 ns 204
              CL30-36-36-76 10.0 ns 178
              • CAS Latency (ns) = CL cycles × (1 / 3 GHz) = CL × 0.333 ns (for DDR5-6000)
              • Sum of primary timings gives a rough idea of overall memory access delays.
              3. Performance Differences
              CL36-44-44-96 vs CL36-36-36-96
              • CAS latency is identical (12.0 ns) at DDR5-6000.
              • The only difference is in secondary timings (tRCD/tRP): 44 vs 36.
              • Real-world difference: Negligible for most workloads, but CL36-36-36-96 may be slightly better for memory-sensitive tasks due to lower secondary timings. In practice, the gain is typically less than 1%.
              CL36-36-36-96 vs CL30-36-36-76
              • CL30-36-36-76 has a lower CAS latency (10.0 ns), which is a 20% reduction in first-word latency compared to CL36 (12.0 ns).
              • Total primary timings are also lower (178 vs 204), indicating generally faster memory access.
              • Real-world difference: CL30 kits can be up to 2–3% faster in memory-bound applications and benchmarks. For gaming and LLM inference, the difference is usually 1–2%—often imperceptible in daily use.
              4. Summary Table
              Timing Set CAS Latency Secondary Timings Real-World Performance
              CL36-44-44-96 12.0 ns tRCD/tRP 44 ≈1% slower than CL36-36-36-96; ≈2% slower than CL30-36-36-76
              CL36-36-36-96 12.0 ns tRCD/tRP 36 Slightly better than CL36-44-44-96; ≈2% slower than CL30-36-36-76
              CL30-36-36-76 10.0 ns tRCD/tRP 36 Best performance; up to 2–3% faster in memory-bound tasks
              5. Recommendation
              • CL36-36-36-96 is preferable to CL36-44-44-96 if prices are similar, but the real-world difference is minimal.
              • CL30-36-36-76 offers the best latency and slightly better performance, but often at a higher price.
              • For your LLM inference workstation, all three kits will perform very similarly; prioritize capacity, stability, and price over small timing differences unless you are optimizing for absolute maximum performance.
              • @ggbhai: Depends what you're after. If you want maximum performance always lean to the tighter timing RAM as not only will it be a few per cent faster if you in the future manually tweak the timings further you could be looking at 4-9% improvement again for CPU bottlenecked workloads/games. Secondary and tertiary timings make the most difference for ddr5, but to be able to do this requires high qualiy Hynix A or M die sticks.

                Though technically speaking Hynix a die doesn't have much impact for ryzen as they work best at higher frequency and amd ryzen is limited to 6400 (and I've personally never gotten 6400 to post stably for my 2x16gb hynix a's). But they will give more leeway on tightening the timings which is where any real performance boost might be ascertained. If interested further see Hardware Unboxed vid "RAM on steroids", bit old now but probably still relevant to larger size RAM.

Login or Join to leave a comment