Crucial Ballistix Sport LT DDR4-3200 Memory
Affordable high-performance desktop RAM
Shoppers of desktop memory are probably very familiar with Crucial, a brand that has been in business since 1996 and became synonymous with their online Memory Advisor tool (originally the Crucial Memory Selector when that was introduced back in 1998). Beyond offering compatible memory adhering to JEDEC standards for home and business machines Crucial has embraced the enthusiast segment, and since 2004 the Ballistix brand has been a competitor in this space.
Today we’re taking a look at new Ballistix memory in the form of a dual-channel DDR4 desktop kit that offers 3200 MT/s speeds out of the box via XMP 2.0, and has the potential to overclock further. Crucial sent along both 16GB and 32GB kits, and we are focusing on the 16GB kit in this review.
Product highlights for the Ballistix Sport LT series from Crucial:
- Speeds start at 2400 MT/s
- Faster speeds and responsiveness than standard DDR4 memory
- Ideal for gamers and performance enthusiasts
- Multi-channel memory architecture maximizes data rates
- Digital camo heat spreader available in white, gray and red
- Easy plug-and-play installation
- Intel XMP 2.0 profiles for easy configuration
- AMD Ryzen Ready
- Optimized for the latest Intel 300 Series platforms
- Limited lifetime warranty
These UDIMMs are part of the Sport LT series, offering a smaller overall footprint while still providing some impressive performance numbers via XMP 2.0 profiles. We tested it out in an Intel system and then moved on to have some fun with memory overclocking in a Ryzen 5 2400G system with integrated Vega graphics. Read on to see how it performed, and if faster memory can make a noticeable difference.
Crucial Ballistix Sport LT 3200 MHz Specifications | |
---|---|
Series | Ballistix Sport |
Model Number | 16GB Kit: BLS2K8G4D32AESBK 32GB Kit: BLS2K16G4D32AESB |
Speed | 3200 MT/s (PC4-25600) |
Timings | 16-18-18 |
Voltage | 1.35V |
Form Factor | UDIMM |
ECC | Non-ECC |
DIMM Type | Unbuffered |
Configuration | 2048Meg x 64 |
Warranty | Limited Lifetime |
Current Pricing and Availability:
- 16GB Kit: $97.49, Amazon.com
- 32GB Kit: $213.99, Amazon.com
The Sport LT modules are available in three colors, with white and red finishes offered as well as the gray version we have in for review. These LT DIMMs offer a lower-profile heatsink that extends only slightly above the standard DIMM height, making them more compatible with various CPU coolers while still offering enhanced heat dissipation to assist with the overclocking potential associated with 1.35V DDR4 memory.
Performance – Intel System
One of the hardest things to do when assessing computer memory is establishing real-world performance gains. Sure, synthetic memory tests can show an immediate and sometimes significant performance uplift from faster speeds and lower latency, but translating this into obvious improvements for everyday use can be difficult. This is not the case with systems relying on integrated graphics which make use of system memory as VRAM, and here you can never have too much memory bandwidth. Increases to memory speed with AMD APUs have been a great benchmark, as faster memory directly translates into higher FPS numbers in games as these iGPUs are bandwidth constrained.
Back when I was first trying to figure out how to review memory five years ago I first attempted to run benchmarks with an intel system before quickly moving on to an AMD platform with the highest-end (at the time) APU. My pair of DDR3 sticks rated at up to 2666 MHz did indeed make a big difference, especially considering DDR3’s standard 1600 MHz speed. I had not re-visited the labourious and generally unexciting world of memory benchmarking until getting these Crucial sticks in for review, but at least I knew going in larger and more useful gains were going to be obvious on the AMD side of things.
On the Intel platform I tested the Ballistix Sport LT modules extensively at their rated speed and voltage using the XMP 2.0 profile, and it was completely stable in normal use and through all benchmark runs.
PC Perspective Intel Z390 Test Platform | |
---|---|
Processor | Intel Core i7-8700K |
Motherboard | GIGABYTE Z390 AORUS PRO |
Memory | Crucial Ballistix Sport LT 16GB (8GBx2) DDR4-3200 |
Graphics Card | NVIDIA GeForce RTX 2080 Founders Edition |
Storage | CORSAIR Neutron XTi 480GB SSD |
Power Supply | CORSAIR RM1000x 1000W |
CPU Cooler | Cooler Master Hyper 212 RGB Black Edition |
Operating System | Windows 10 64-bit |
To attempt to assess performance we begin with some preliminary results using AIDA64 to demonstrate the throughput of the memory beginning with its default 2400 MT/s speed and ranging up to the highest stable overclock I could get without adding much voltage past 1.35V.
The memory timings begin with 16-16-16-39 at DDR-2400, with the XMP 2.0 profile offering 16-18-18-36 at DDR-3200. I attempted to improve on the XMP timings at 3200 MT/s and was able to tighten thse up a little when increasing the voltage. I also found DDR4-3600 to be easily reached, with timings of 18-18-18-40, after a voltage increase to ~1.370V (this board offers memory adjustment in .010V increments); and with some additional work/higher voltage I could have hit some better numbers.
As expected in this synthetic benchmark the performance scaled perfectly as memory bandwidth increased. So how to put this to good use? There are certainly applications that can take advantage of higher memory clocks, such as 7-Zip. Rather than use the built-in benchmark tool I decided to create a folder of exactly 5GB filled with an assortment of file types, and then compressed it using .7z at the maximum setting.
From the stock 2400 to the XMP profile at 3200 there was a significant jump, with the total time dropping over 30 seconds just in this short compression test. Compression/decompression are memory intensive so fast RAM will help cut down on some pretty significant time here even in this short test scenario. What follows is an example of very minor improvements from memory speed, as I found when experimenting with the latest version of the HWBot x265 benchmark.
While barely perceptible the gains in the brief 4K benchmark are measurable and would translate to some time savings in a longer encoding tasks, and in future it may prove more useful to time a typical encoding task using Handbrake.
There were no surprises with these quick Intel platform benchmarks, and one of the hardest things to do when assessing computer memory (on Intel systems anyhow) is establishing real-world performance gains. Yes, synthetic memory tests can – and did – show an immediate and sometimes significant performance uplift from faster speeds and lower latency, but translating this into obvious improvements for everyday use can be tricky (and time-consuming).
It makes a lot of sense to focus on the AMD side for system performance, as Ryzen CPUs are well known for benefiting from increases in memory bandwidth, and going forward we can do just that. For now, however, we will take a quick look at gaming performance with integrated graphics using a Ryzen 5 2400G.
Performance – AMD System with iGPU
I think we all know at this point about the importance of fast memory with Ryzen processors, with Infinity Fabric relying on RAM bandwidth for faster CPU performance across CCX units, and to this end more benchmarking will be done with the next RAM review. For this review, however, I concentrated on validating the 3200 MHz speed on the AMD B450 platform, and the benchmarks on the AMD side are limited to my experiments with a Ryzen 5 2400G processor which offers Vega graphics onboard. Compared to the Intel side of things benchmarking memory is easy with systems relying on integrated graphics, as these of course make use of system memory as VRAM. In this scenario you can never have too much memory bandwidth and increases to memory speed with AMD APUs have been a great benchmark in the past, offering a compelling real-world example of what faster RAM is capable of.
So using an AMD B450 system equipped with an AMD Ryzen 5 2400G with integrated Vega 11 graphics it was time to test some 1920×1080 gaming. You may notice that voltage was a little higher with the overclocked memory here compared to the Intel system, and for whatever reason I need to push things to 1.380V with this test setup. In the past I have had to run slightly higher voltages for stable RAM overclocks on AMD platforms, but that could be coincidental.
PC Perspective AMD B450 Test Platform | |
---|---|
Processor | AMD Ryzen 5 2400G |
Motherboard | ASUS ROG Strix B450-I Gaming |
Memory | Crucial Ballistix Sport LT 16GB (8GBx2) DDR4-3200 |
Graphics Card | Integrated RX Vega 11 Graphics |
Storage | CORSAIR Neutron XTi 480GB SSD |
Power Supply | CORSAIR RM1000x 1000W |
CPU Cooler | be quiet! Dark Rock 4 |
Operating System | Windows 10 64-bit |
I tested just a couple of games here, with Civilization VI known to use a mix of CPU and GPU, and Ashes of the Singularity Escalation (using the Vulkan API) is another game that provides a good test of the overall system.
In Civ 6 the average FPS improved from 28.96 with the default 2400 MHz speed to 31.69 at the XMP equivalent, moving up to 32.44 with the memory overclocked further to 3466 MHz.
In Ashes there was a more significant jump from the default 2400 MHz to 3200 MHz, with averages moving from 37.67 to 42.49 FPS. The 3466 MHz overclock presented a lower than expected boost, with an average of 42.83 FPS.
Yes, faster memory can directly translate into higher FPS numbers in games as this RX Vega 11 iGPU is quite obviously bandwidth-constrained. It's worth noting that with Ashes of the Singularity: Escalation in particular the lower latency of the XMP-equivalent settings provided more consistent frame times, as the max stable 3466 MHz OC's looser timings offered very little over the 3200 MT/s rating of this memory with its solid 16-18-18-36 XMP timings at that speed.
Conclusion
Crucial's Ballistix Sport LT memory is priced competitively, offers excellent performance out of the box with XMP profiles that provide very good speeds and timings with zero effort, and these DDR4-3200 kits are a fine addition to the series. As someone who has upgraded and supported many OEM systems over the years I have been personally buying Crucial memory for a long time, and the Ballistix brand – while 15 years old itself at this point – is perhaps something of a sleeper in the high-performance RAM market. If you compare specs on memory you might be surprised at how competitive Ballistix memory tends to be, and this model in particular at 3200 MT/s with 16-18-18-36 is an excellent performer and a great value at its current pricing.
Beyond its default settings I also found this Ballistix Sport LT 3200 memory to overclock up to 3466 MHz on the AMD platform and 3600 MHz on the Intel system without any trouble, and more adventurous overclockers will likely find better timings than I did at these speeds. And whether you choose to use the XMP profile or try overclocking yourself, these are easily recommended as they provided reliable performance through all testing.
“AMD Ryzen Ready”
Is it a
“AMD Ryzen Ready”
Is it a joke? 😀
“•AMD Ryzen Ready”
Well
“•AMD Ryzen Ready”
Well that’s not so bad to be tested/validated for use with Ryzen SKUs for folks building AMD based systems.
Article even states:
“•Optimized for the latest Intel 300 Series platforms”
So that’s good for Intel usage also.
The memory timings are not the tightest but the pricing is not as high also and a little higher stable memory clocks for any current Zen/Zen+ based system is good for the Infinity Fabric also.
Maybe as the article states there can be a little more memory tweaking and even better results for AMD and I’d like to see the SO-DIMM variant tested if available for that DeskMini A300 Barebones Mini-STX(1) PC for AM4 SKU.
I really would like to see more websites source the ASRock DeskMini A300 Barebones Mini-STX(1) PC for AM4 simply because it’s the only current STX/Barebones option for AMD’s APUs. It’s so that the STX form factor can be a test bed for all sort of SO-DIMM Memory SKUs and readers able to see what is the better memory for that STX form factor platform. They could also source the Intel based variant(ASRock 310 Series DeskMini) and also use that as a STX based form factor test bed for Memory testing also on low cost Intel barebones/STX systems.
So considering how popular the Mini STX/other mini form factor offerings are currently I’d always want to know if there are SO-DIMM variants avilable also. That and use the ASRock(300/310) paltform variants for benchmarking Intel CPUs/Integrated Graphics against AMD’s APUs/Integrated graphics on a very similar in design/form factor STX Platform ASRock offering. That case looks the same between the 300(AMD) and the 310(Intel) ASRock DeskMini variants so that makes these very good candidates to become STX testing platforms to test AMD’s APU/Integrated Vega Graphics against Intel’s CPU/Integrated Graphics in a Mini-STX form factor.
(1)
“ASRock Launches DeskMini A300 Barebones Mini-STX PC Supporting AMD CPUs”
https://www.pcper.com/news/General-Tech/ASRock-Launches-DeskMini-A300-Barebones-Mini-STX-PC-Supporting-AMD-CPUs
“Well that’s not so bad to be
“Well that’s not so bad to be tested/validated for use with Ryzen SKUs for folks building AMD based systems.”
Validated for 1.35 V but won’t work out-of-the-box for Ryzen processors without rising the voltage to 1.38 V.
Well they should rather market it as “AMD Ryzen’s Incompatibility Ready”! :o)
If so then maybe that’s a job
If so then maybe that’s a job for the FTC, but it all depends on how Crutial/The Industry markets their RAM/RAM overclocking on their memory products. This is for overclocking outside of the JEDEC specifications so maybe the voltage needing to be a little higher for AMD is no big deal.
The Stock profiles and XMP(AMD calls is AMP) profiles are two different things but XMP/AMP are outside of the JEDEC standards for memory usage and that overclocking technology has been in usage for some time. The article uses XMP(Intel IP) naming but the MB makers all have access to that same sort of IP even for AMD’s AM4/TR4 MBs.
As far as I can tell the if the MB maker os saying 3200 MT/s (PC4-25600) at 1.35V then most of these SKUs better clock to 3200 MT/s with no issues and if they are not stable at that rate and at that voltage then that may be cause for a valid RMA.
So you may just be correct but all this XMP/AMP usage is outside of the JEDEC standard usage anyways and AMD’s memory controllers in CPUs/APUs are still not as performant in the memory subsystems department compared to Intel. So AMD’s Zen First Generation Raven Ridge 2000 series APUs on GF’s 14nm node require a little more voltage than Intel and now I’m wanting to see what the Picasso Desktop Zen+ RR APUs(3000 series) on GF’s 12nm node with some memory controller tweaks will do with this memory.
I did say tested/validated and assuming that because that’s what it should be to make that claim of 1.35V but yes you are correct that memory maker should not be stating 1.35 at 3200 MT/s unless 99.8% of their samples are only needing 1.35V stable. The Term “Ryzen Ready” probably means at least for the inside the JEDEC standards usage the memory will at least boot up and work for Ryzen.
But everybody Knows that First Generation Ryzen had some memory overclocking issues and even the First Generation Raven Ridge APU’s memory controllers should not perform as well as the Raven Ridge Picasso Zen+/12nm variants in the memory controller department.
Now I really want to see just how any Raven Ridge 3000 series Desktop APUs will perform with this memory but so far only the mobile Raven Ridge 3000 series SKUs are currently available in a few OEM/Laptop SKUs.
By the way where the hell are the Zen+ Desktop Picasso 12nm APU variants, or even news of any RTM ETA and product line up for Desktop 3000/Picasso series APUs?
“The Term “Ryzen Ready”
“The Term “Ryzen Ready” probably means at least for the inside the JEDEC standards usage the memory will at least boot up and work for Ryzen.”
To be “AMD Ryzen Ready” they should at least provide profiles to reach 2933 MT/s with MEMCLK = 1467 MHz and 2666 MT/s with MEMCLK = 1333 MHz.
see:
https://www.reddit.com/r/Amd/comments/b4q01o/a_quick_analysis_of_the_new_zen_2_bios_options/
If your Ryzen can’t reach its full specifications then he’s NOT “Ready”.
To support all configurations, memory module manufacturers have to provide SPD EEPROM over 256 bytes with compatible BIOS. It’s likely cheaper (cf. logistic cost) than producing specific modules for low volume AMD chips.
see:
https://en.wikipedia.org/wiki/Extreme_Memory_Profile
Well AMD, Intel/others, and
Well AMD, Intel/others, and the memory makers are really the entities that make up JEDEC and JEDEC’s working committees and JEDEC does not classify any of the unused bytes that are in there but that are now being used for XMP/AMP.
JEDEC standards and XMP are not one and the same:
“A similar, Intel-developed JEDEC SPD extension for DDR3 SDRAM DIMMs, later used in DDR4 also. XMP uses bytes 176–255, which are unallocated by JEDEC, to encode higher-performance memory timings.[15]” (From That Wikipedia Extreme_Memory_Profile entry that you refrenced]
This is low cost memory and folks know that fact and JEDEC standards and the XMP extention created by Intel are not going to be seen as an enforced requirement for any makers that can only be required to meet the minimal JEDEC standards that are not based on any CPU maker’s extentions.
“Ryzen Ready” can be the lowest range of memory speeds and timings officially supported by the JEDEC standard and any consumer usage beyond the minimal JEDEC standards can be considered to be the consumer’s responsibility.
If you are using the Ryzen 5 2400G and get this memory and are not able to be satisified with its performance then return the memory by the seller’s alotted return timeline. But to expect any 14nm first generation Zen CPU or APU parts to have the best memory metrics on all memory kits is maybe being a little unrealistic.
The First generation AMD consumer CPU/APU parts are not as well binned for memory usage comapred to the Server graded/binned parts and the consumer market overclocks memory outside of the JEDEC standards using Intel Invented XMP extentions so that’s mostly on the buyer to decide in the end if they can not get the performance that they need from any memory kit.
Most consumers will not be wanting to overspend when they are only using an AMD Ryzen 5 2400G, First Generation, APU part that can be had at sale pricing. So many will at least try this memory in hopes that their APU sample is able to perform better than the one that PCPer used in their review. And that PCper reviewed SKU sample does not represent a large enough of a APU sample size with which to judge any CPU’s/Memory Kit’s memory performance.
Remenber that before Intel turned Overclocking into that “K” market segement to charge more for that new Intel “K” market segement that original overclocking was not about getting any top end part that was specifically binned/created for Overclocking(K series). Overclocking was originally about getting the lower binned part where all parts are unlocked mostly and getting that lower binned part for less money and making that part perform similar to the higher binned part that costs more money.
So for AMD all the parts are mostly unlocked and some overclockers were getting the lower binned Ryzen parts, Ryzen 7 1700 for example, and getting that part to perform more like the Ryzen 7 1800, and ditto for the other Ryzen binns across 2 generations of Ryzen, soon to be 3 generations, SKUs.
The PCPer article gets the memory up to 3466Mhz with looser timings and 3200Mhz with tighter timings and the game perfomance is about the same. So there appears to be diminishing returns going beyond 3200Mhz without using tighter timings and this memory is low cost so that can be expected.
You appear to be expecting too much at this price point and on first generation Raven Ridge on the GF/Samsung 14nm process node but I do hope that when any Ryzen/Raven Ridge 5 “3400G” SKU arrives that this memory will be retested.
Maybe you should contact Crucial and ask them what “Ryzen Ready” implies and report back on their reply. A quick Google search of “Ryzen Ready” appears to mean that the BIOS can handle the CPU without needing to be updated so that probably means a similar thing in that the memory will at least boot and be stable for out of the box usage at the minimum on Ryzen systems. We are talking about some nebulous marketing term when “Ryzen Ready” is used for memory unless AMD has defined that fruther.
But do try and ask the memory makers about “Ryzen Ready” and report back their replies.
“The Stock profiles and
“The Stock profiles and XMP(AMD calls is AMP) profiles are two different things but XMP/AMP are outside of the JEDEC standards for memory usage and that overclocking technology has been in usage for some time.”
You’re right. XMP is Intel’s memory profile specific but AMP didn’t get enough traction among manufacturers.
“The article uses XMP(Intel IP) naming but the MB makers all have access to that same sort of IP even for AMD’s AM4/TR4 MBs.”
AMD should enforce its AMP IP via marketing labels.
I’m not sure if validated “AMD Ryzen Ready” memory modules are effectively tested by the manufacturers.
It seems too easy to market it as “AMD Ryzen Ready” in order to sell low volume AMD modules assuming a low rate of client returns.
It could hurt the Ryzen platform!
AMD is using AMP because
AMD is using AMP because Intel has the “XMP” trademarked but AMP is essentially just XMP anyways as the memory makers are mostly supporting the maker with the largest market share for many years and currently.
It should be on AMD to have to define further what the term “Ryzen Ready” defines when used in the context of memory.
But Overclocking is what it is and that’s outside of the JEDEC standards so that’s more between the consumer and the respective memory maker as to what the consumer will accept. The consumer is the one doing the overclocking on their specific CPU/APU based system so that’s up to the consumer to decide if they are satisified or not with that memory.
There is a lot of Silicon Variance as a result of the diffusion process and the statistical process used by each CPU maker for their respective binning processes! So that’s the usual Silicon Lottery where some consumers are just lucky compared to others. And AMD does make use of more voltage to get better yield/binning numbers so that’s more variance as far as stable voltages are concerned on AMD parts.
AMP is just what it’s legal for AMD to call essentially what the MB markers are doing with Intel’s XMP across both Intel and AMD platforms. Intel got JEDEC to adopt that XMP extention so it’s an open IP for others to revrese engineer if they even have to do that they can most of the info from JEDEC on any extentions and how they work or are defined.
So what AMD can only legally refer to as AMP(Intel has XMP trademarked) is if fact mostly the exact same IP via that JEDEC extention adoption of Intel’s XMP. Intel is also adopting VESA Display Port Adaptive Sync and that’s the same thing as first generation FreeSync(TM) by AMD.
And we see how Industry Standards bodies(JEDEC/others) work and how closed proprietary standards become Open Industry standards over time. USB-IF’s USB 4/4.0 standard is adopting TB3 into its open indusrty USB standards and that’s also how most things work via these Open Standards Bodies that the respetive indusrty creates. Ditto for HBM2/other standards.
Great review thanks. Have
Great review thanks. Have Crucial improved their memory since the old days. I used to have Crucial DDR2 1066MHz and paid big bucks for it because it had a speed rating above the main stream 800MHz. Anyways I had a setup with a Core2Quad Q6600@4.0GHz OC. The memory just seemed to do random problems about every 3 months on the ASUS ROG Maximus board I had at the time. I would have to re seat the memory every 3 months and it would be fine again for 3 months. I contacted Asus about this and they said yes Crucial memory and that board sometimes could have problems.
I changed the memory out to some cheap Kingston DDR2 800Mhz ran it at 1066MHz and never had a problem again. Actually I still use the system at work with the same Kingston memory today and it has never had any farther problems since taking out the Crucial ram. I tried everything with the Crucial before finally removing it even tried 800MHz no joy as well.
over the years I tried the Crucial sticks in several PC’s and would run into problems. I did finally install it into a AMD board with a AMD dual core and 800MHz memory speed and it finally behaved as it should so the memory was fine just something weird in their setup. Once I deemed the system to be stable with that memory I finally rid myself of those Crucial sticks by selling the system.
Long story short it will be a very long time before I would be willing to trust Crucial memory ever again.
Because you got 1 bad stick
Because you got 1 bad stick in 1996, you’ve sworn it off huh? Well they produce good ram, 20 million systems since then have run it fine.
1 bad stick?? Both sticks
1 bad stick?? Both sticks were fine. I said that on my ASUS board and others I tried them on after I finally gave up on them because of not being fully compatible with a lot of boards in that time frame. They did finally work fine on a Intel board I tried them on in a system I built to sell a couple years later.
It was not 1996 more like 2007 when the Q6600 was released.
Nice review, but hook a
Nice review, but hook a brutha up with some multi-threading benchmarks, like wut’s the performance like when you computer is up to a bunch of bidness, since ya know, i got a bunch of cores now and i often do more than one thang, know what i’m saiyan?
Them cores are hangry man, need lots of memory accesses dog when the cache runs out, CREAM Cashe Rules Everything Around Me, until it runs out, and I gotta get off my ass an run to a bank or some shite, like a computer reaching out to next level memory, nawmean?
Anyways, I be kickinit with da bling bling down in the Hamptons, gotta date wid a big tittied cougar ina minute. Plannin’ to stuff all my cache in her spching spching nawmean? See ya’ll on the flip side.
Yo, this CAPTCHA shite is about to make me abandon this site for good. I use TOR, and I ain’t gonna quit usin’ it for yuus gize.