Introduction, Specifications, and Pricing
While Silicon Power should pretty much be a household name at this point, no matter your storage needs, yet in the time we have been reviewing RAM, this is but the thirds set to hit the office. We know that what we have seen in the past is a bit of a mixed bag. Our first kit was something we feel would have been more at home in a G.Skill Ripjaws lineup, but even with the limited speed of the XMP profile, there was some movement to fine-tune their performance. The second kit we had was very similar in the basic style, but this time the Xpower Turbines came with RGB lighting, and while they looked different from the first set of Turbine RAM we had, they were nearly identical under the hood.
We have a slightly faster set of DDR4 and twice the density on the same number of sticks. With that increase in speed comes looser timings, but it appears Silicon Power may have run out of their Samsung IC surplus this time around. And we find this set to be based on Hynix ICs. While they were changing ICs, we were also dealing with a different series from Silicon Power. What you are going to see are not the familiar-looking Xpower heat spreaders, but something completely different that we do not believe has been done before when it comes to styling and shape.
As we first look at the Xpower Zenith RGB kits offered from Silicon Power, we will say this much. Even if you do not appreciate the looks at first glance, those who may sneer at the use of Hynix ICs may be in for a bit of a shock. Not only can Corsair deliver a Swan Song for their last drop to the DDR4 market, but it appears Silicon Power is also firing on all cylinders and is well worth the time it takes to see just how well they did.
The specifications we show were taken directly from the product page, and the first things we see are images of the RAM stick with dimensions offered. At 38.5mm of height, we see that this kit will sit under almost any air cooler without issue. The 133.4mm is standard for length, but we like that Silicon Power also offers the thickness, as it could play up with a CPU cooler offset when all slots are full.
The rest of what is shows starts with a display of densities. The Xpower Zenith RGB can be had as small as a single 8GB stick, and there are 16GB and 32GB sticks, which sends the max density for these to 64GB over two sticks. It also shows that there are three steps in speed bins as well. The basic kits start at 3200 MHz, which isn't all that bad, and the mid-range kits are offered at 3600 MHz, both using 1.35V. The top-tier offering is the 4133 MHz kits using 1.4V to obtain that speed. Timings also depend on which kit you pick, where the slower kit is set for CAS16, the Xpower Zenith RGB we have at 18-22-22-42, or CAS19 for the faster stuff. The last thing we see is that Silicon Power offers a lifetime warranty, which is nearly the industry standard.
With a fresh set of SP032GXLZU360BDD in our hands and a set of thirty-two Hynix ICs in a dual-sided arrangement, which should help with performance, much like we saw in the Vengeance RGB RT, possibly even better! While Silicon Power whitewashed the ICs, with what this kit can, they could have micro-SD cards glued to the PCB, and we could care less. The out of the box performance, along with a bunch of wiggle room for overclockers, is what is going to carry this kit into customers' machines.
With all of this sort of talk, we may as well jump right in and get to the point, as you may not know Silicon Power is capable of something this good, but with what follows, there should be no question that they are serious about giving customers what they want.
Packaging and Silicon Power Zenith RGB
The packaging of the Xpower Zenith RGB is typical for Silicon Power, where they use clamshell packaging made of clear plastic, with a cardboard insert adding style and color to the package. While we can see both of the RAM sticks, we need to look between them for the details of this kit. We can see it is Zenith RGB, but we cannot forget the Xpower moniker on the heat spreaders. Moving on, we then see icons for the various supported sync options and see that this is a 32GB kit of DDR4.
The back of the packaging looks at the back of the cardboard insert through the clamshell packaging. On it, we see the company name and the QR-code at the top, followed by a statement that this RAM supports the latest XMP standards, with guaranteed compatibility. Sadly there is no mention of Ryzen support, but we will find out how that goes anyways. Beyond the multiple versions of that statement, there is information on obtaining support should you need some. We also see the part number on display in the lower right corner.
Now unencumbered by the plastic, we get our first clear look at the Xpower Zenioth RGB in all of its glory. Visually we have something new, not with the brushed aluminum texture, not with the dark gray color either. However, the curvy top portion and the basketweave look on the sides of the spreader is something not done before. Beyond that, the black PCB was a must, and we have no issue with the painted on XPOWER or the DDR4 Gaming Memory, which is done in white for high contrast.
We like that Silicon Power makes both sides of the heat spreader identical in shape, color, and text painted on it. On the far left end of these sticks is where you will find the colorful product sticker. You will find the stick part number on it, not that of the kit, but we do see the speed, CAS latency, and density of the stick all offered on it, along with the serial number at the bottom.
We like this angle for the image, as it shows the depth of the pattern much better than when they are laid flat on the table. Even though the basic perimeter shape is "standard," we do like the originality of this design.
We also like the reversal of roles when we see the top of this RAM. As the sides were dark with bright lettering, the white light bar is painted with black text that mimics what we saw on the sides. The Styling carries into the light bar, with it being much fatter in the middle than at either end.
With the bulk of the images done, we had a look at what ICs were under the hood. The speed grade is 2666V downbin, the organization is dual-rank, and we are shown these to be on 8-layer B1 PCBs. The right half shows us that these are Hynix but delivers only a partial H5AN8G8N??R-VKC, but we would hazard to guess it is C-die.
With there still being a question about the exact ICs, we popped the hood manually and looked for ourselves. Running the numbers and letters through Chrome lead us to no answers, as they only mean something to Silicon Power. On the right edge of the PCB, we can also see the LEDs, which are spread a tad too far apart for our liking.
Once we had the Xpower Zenith RGB installed to be controlled with our 3900X, we booted the system and took this image to show what Silicon Power offers. We have no issue with how well the heat spreaders blend into the build, and to be honest, the sticker showing on the side is not great but is not the main topic at this time. What is, is the bright hotspots in the light bar from using just 5 LEDs per stick. The colors are bright and presented well, but we do wish it were a smoother blend.
Paired with the 10700K, the RAM sits right next to each other on the Apex and almost makes the bright pots worse, or at least more noticeable, in our opinion. Not to harp on the lighting continually, but at the end of it all, this is the one thing that may make or break your decision when it comes to buying these or something else equally priced.
Test System Details
To obtain the AMD CPU-Z screenshots, you will see directly following this image, and this is the system we used to do it, as well as in attaining the results seen in the following pages. Thanks go out to Corsair, ASUS, and GIGABYTE for supporting this venture. Detailed specifications of the system can be found below.
Those who stick this kit in a motherboard and run it as-is may help to know that the SPD profile is set at 2666 MHz with 20-29-29-43 1T timings at 1.20V. Those willing to take the trip into the UEFI will enable DOCP, which you can see delivers the specified speed and timings. 3600 MHz with 18-22-22-42 1T using 1.35V is what we expected and is what we got. If you wonder about the SOC voltage, it is 1.016V with the SPD profile and 1.080V with DOCP enabled.
Doing what we do, we attempted to reduce the timings. In that pursuit, we ended up with 15-22-22-42 1T with the VDIMM raised to 1.45V and the SOC got a tick added to it where it now is at 1.18125V. Attempting CAS14 resulted in postcode errors of 09 and 00, and any time we tried to move the secondary timings, we were greeted with past code 22.
To be blunt, with as many ICs at play here and the fact that timings did not move much, we were impressed with the 400 MHz bump left in the tank. We are using the DOCP profile and raising the RAM speed, still using the same voltages mentioned in the timings portion. 4000 MHz at 18-22-22-42 1T is still better than what many kits spec'd at that speed come with as far as timings go.
Chad's AMD DDR4 Dual-Channel Test System Specifications
- Motherboard: ASUS Crosshair VIII HERO Wi-Fi - Buy from Amazon
- CPU: AMD Ryzen 9 3900X - Buy from Amazon
- Cooler: Corsair H150i PRO - Buy from Amazon
- Video Card: GIGABYTE GeForce RTX 2060 SUPER Gaming OC 8GB - Buy from Amazon
- Storage: Corsair Force MP500 480GB NVMe - Buy from Amazon
- Case: Thermaltake Core P5 TG - Buy from Amazon
- Power Supply: Corsair RM750x 750-watt - Buy from Amazon
- OS: Microsoft Windows 10 Home 64-bit - Buy from Amazon
To obtain the Intel CPU-Z screenshots, you will see directly following this image, and this is the system we used to do it, as well as in attaining the results seen in the following pages. Thanks go out to Corsair, ASUS, and GIGABYTE for supporting us here too! Detailed specifications of the system can be found below.
Swapping to the Intel rig, we again visited the UEFI and enabled the XMP option. Doing so delivers us the Xpower Zenith RGB in the manner we were told it would operate. 3600 MHz with 18-22-22-42 2T timings is what we expected. The VDIMM is 1.35V, the VCCIO is 1.312V, and the VCCSA is 1.504V
We do not feel that the difference here is based on the IMCs, but rather that we moved to 2T, which allows us to reduce timings a bit further. Changing only the VDIMM to 1.45V, we concluded that 15-19-19-42 2T is the bottom of the range for this RAM. Trying 15-18-18 results in a boot loop, and CAS 14 resulted in different postcodes.
Our last venture into tuning results in what CPU-Z shows us. We passed what was possible with the AMD system, raising the bar from 4000 MHz to 4133 MHz on the Apex. To do this, we raised the VDIMM to 1,45V, leaving the IO and SA voltage where they started. We could post into Windows as high as 4266 MHz, but we could not pass testing.
Chad's Intel DDR4 Dual-Channel Test System Specifications
Chad's Intel DDR4 Dual-Channel Test System Specifications
- Motherboard: ASUS ROG Maximus XII Apex - Buy from Amazon
- CPU: Intel Core i7 10700K - Buy from Amazon
- Cooler: Corsair iCUE H150i Elite Capellix - Buy from Amazon
- Video Card: GIGABYTE GeForce RTX 2060 SUPER Gaming OC 8GB - Buy from Amazon
- Storage: Intel SSDPEK1W120GA
- Case: Thermaltake Core P3 - Buy from Amazon
- Power Supply: Corsair RM750 - Buy from Amazon
- OS: Microsoft Windows 10 Home 64-bit - Buy from Amazon
Right out of the box, the Aida read performance is not bad at all, resulting in that 54495 MB/s, delivering Silicon Power to a fourth-place finish. We did find there to be a 75 MB/s deficit for running with tighter timings, but that top result of 56105 MB/s is the fasted of any kit tested on this system.
Write performance shows Silicon Power in third place this time, with the DOCP results totaling 53199 MB/s. We again to a penalty for lower timings to the tune of 45 MB/s. However, that 56797 MB/s from this kit with more speed is the best we have seen since getting this setup.
Silicon Power is back in fourth place when it comes to copy performance with 49907 MB/s. This time, we took advantage of the tighter timings, as the throughput increased by 425 MB/s. If you would like another 3600+ MB/s in performance, it is there to be had if you are willing to push the speed a bit.
Latency is not horrible either. With a fifth-place finish, DOCP delivers a 73ns result. We lowered that to 70.7 by running them at 3600 MHz CAS15, but we took a decent-sized hit when we raised the speed, ending up at 80.8ns for that run.
Fire Strike Physics scores do not scale well by speed and primary timings alone, and with that said, we are pleased with their second-place finish while running DOCP. While both options for clocking this kit showed an advantage, in this instance, speed wins out over tighter timings.
As PCMark 10 tends to do, it threw Silicon Power a curveball, landing the Xpower Zenith RGB in eleventh place overall. While we wholeheartedly expected speed to help, it doesn't, but running them at CAS15 bumps them into sixth place.
We cannot imagine how it gets better than these results when using 7-zip to compress data. In our short run, the Zenith RGB with DOCP enabled beats the next kit in line by more than eleven seconds! Speed did not help, but the penalty is not bad at all, but when you look at that sub-400 second CAS15 run, there are no words from a score that only the Corsair Dominator Platinum RGB were able to do when they were overclocked.
If Cinebench is your go-to metric for what RAM you will buy, you may want to look past this chart. Out-of-the-box performance is dismal, landing the Xpower Zenith RGB in last place. Both options for overclocking improved things but was still not enough to make it past seventh in the chart.
Transcoding with Handbrake is something many do, especially anyone who does video production. Using DOCP, Silicon Power lands in third place, trying to keep up with the Vengeance RGB RT. More speed netted us a near eleven-second reduction in the time to complete, but we gained another seven seconds by running them at 15-22-22-42 1T.
Now dealing with the Intel performance, we enabled XMP and ran AIDA64 to see what shakes. The Silicon Power Xpower Zenith RGB placed seventh overall, with the 49079 MB/s being nearly 500 MB/s slower than the Vengeance RGB RT. With a latency drop, we did improve to 50001 MB/s gaining a couple of places, but if you are looking for the best results, speed is where it is at, delivering us with 54912 MB/s.
Third-place is an excellent place to start with the XMP figure of 53748 MB/s. Running this kit at CAS15 does have an advantage of roughly 150 MB/s, but that 61398 MB/s from increasing speed is the best result to date, some 3000 MB/s better than the Vengeance RGB RT delivered.
The Copy performance shows similar results, but now Silicon Power is in second place at 47436 MB/s. Reducing timings gains us almost 1200 MB/s more, but that 53943 we see at the top of the chart is again the best we have seen to date, with the 54943 MB/s result. Besting the next in line, the Vengeance RGB RT overclocked runs by some 3355 MB/s.
The 54.2ns latency is seventh from the bottom of the chart, but it is still slightly above average. However, when we opted for CAS15, we got down to 50.9ns, and with this kit, at 4133 MHz, we could lower it a bit more.
To say that these results were not a surprise is an understatement. We know that Fire Strike Physics scores are tough to get, but the Zenith RGB finds itself at the top of the chart out of the box. More speed results in a slightly lower score, although still good, but trying them at 15-19-19-42 results in a trip down the chart to ninth-place.
PCMark 10 is a benchmark that has handed more than one kit of RAM its ass on a silver platter, but Silicon Power does not run into this at all! Taking top honors with the USE of XMP is fantastic, and the fact that we can linearly increase the score is also a bit of a shock.
Again, Silicon Power comes out swinging for a great result in compression of files. The top spot at 449.984 seconds is excellent, but we can pull another twenty seconds from that XMP run by lowering the timings. On top of that, we can gain another eight, nearly nine seconds, by increasing speed.
Cinebench results are not a shocker, as it tends not to play nice with the kits you would expect. However, even though the XMP results are fifth from last, things can be improved. There is nothing extraordinary as far as chart-topping scores here, but it is nice to see that it does not flat line near the bottom of this chart.
Handbrake testing does not fare well for the Xpower Zenith RGB, which is fifth from last again. However, if you do not mind playing around in the UEFI and doing some testing, you can shoot those results to very respectable levels, chart-topping even.
When we begin to discuss performance, we look back at the charts and see that when it comes to AIDA64, these Silicon Power sticks do well compared to the rest of the field. On the AMD system, this sort of performance was found in all but two tests we do, where we find that the Xpower Zenith RGB is not only better than average, it excels at placing much closer to the top of our charts than they do the middle. In the Intel camp, where these sticks are tuned to run their best, we find that not to be the right way to look at it.
There are three or four benchmarks where Silicon Power did not perform the way we expected but is mimicked with the AMD results on a couple of them. Not bad out of the box, but this is a kit for the tinkerers. There is a bunch of flexibility left to deliver some of the best numbers we have seen to date in any DDR4 tested on this pair of systems, and that will come into play with the scoring and our overall opinion, even without there being a guarantee of this type of movement within that many ICs.
Looks are personal, and while we do not mind the heat spreaders or the overall theme in any way, we do feel that the RGB LED lighting is what will have many opting for something else. It is blunt of us to say this but at a similar cost. Compared to the Vengeance RGB RT we just looked at with a fantastic display of RGB LEDs, these hotspots will turn some away. The good thing about it all is this, the only complaint we have about what Silicon Power provided us is the lighting.
It is not that Silicon Power did something inherently wrong here. It is there are better-looking lighting options on the market from just about everyone. Yet, even with this complaint, we like the basketweave design and applaud Silicon Power for doing something different, something we have yet to see.
Performance is the top priority, and Silicon Power did impress us in that respect. For those of you who do like what you saw here, whether it be performance-based or an aesthetic choice, you are going to be hard-pressed to find them on this side of the pond. We were able to find one seller, although we do not know who they are. They were selling kits at $218.51, which places the Xpower Zenith RGB in a head-to-head battle with the Vengeance RGB RT.
We do know that the price is slightly more than the $203.20 MSRP, but even there, the bar is raised too high to where Silicon Power should have done more, in our opinion. At this point, it is just a matter of what is readily available, and sadly it does not appear Silicon Power is trying to win that battle either. We feel they may have priced themselves out of the market even before we spend way too much time not finding them for sale.
The Bottom Line
The Xpower Zenith RGB is quite the kit, delivering performance that surpasses all of our tested kits at this time. Even though it looks good, the lighting is a bit of a downer, and the sheer lack of obtaining them is likely to have you looking elsewhere.