Benchmarks, Overclocking And DDR3-v-DDR4 Evaluation
Before we could start testing, we had to make sure the comparison is fair. The fact that DDR3-2800 is priced similarly to DDR4-3333 indicates that a comparison to DDR4-2800 wouldn't be equitable. For any given market segment, DDR4 will have a higher data rate and worse timings compared to DDR3.
But are the timings that much worse? Consider that CAS 11 is the most-common latency standard for DDR3-1600, and that CAS 15 is the most-common latency standard for DDR4-2133. A 1GHz cycle takes one nanosecond, and simple math tells us that 1 / 2.133 * 15 = 7ns, while 1 / 1.600 * 11 = 6.875ns. Mathematically, that's not much of an advantage.
Klevv's modules are rated at DDR3-2800 CAS 14 and DDR4-3000 CAS 16, but ASRock's low-cost boards can't hold those data rates. Limiting ourselves to default CAS 11 (DDR3) at 1.50V DIMM and CAS 15 (DDR4) at 1.30V, we found we could still reach DDR3-2400 and DDR4-2666. That gives the DDR3 modules even more of a latency advantage at 4.58ns, compared to DDR4 at 5.625ns. Had we worked any harder to put DDR3 in a good light, DDR4 guys would rightly cry foul.
Test System Configuration
|Chipset||Intel INF 10.0.27|
While the rest of the testbed is carried over from our previous Z170 motherboard reviews, the change in memory for today's comparison required a complete retest.
|Battlefield 4||Version 22.214.171.124, DirectX 11, 100-sec. Fraps "Tashgar" Test Set 1: Medium Quality Preset, No AA, 4X AF, SSAO Test Set 2: Ultra Quality Preset, 4X MSAA, 16X AF, HBAO|
|Grid 2||Version 126.96.36.19979, Direct X 11, Built-in Benchmark Test Set 1: High Quality, No AA Test Set 2: Ultra Quality, 8x MSAA|
|Arma 3||Version 1.08.113494, 30-Sec. Fraps "Infantry Showcase" Test Set 1: Standard Preset, No AA, Standard AF Test Set 2: Ultra Preset, 8x FSAA, Ultra AF|
|Far Cry 3||V. 1.04, DirectX 11, 50-sec. Fraps "Amanaki Outpost" Test Set 1: High Quality, No AA, Standard ATC, SSAO Test Set 2: Ultra Quality, 4x MSAA, Enhanced ATC, HDAO|
|Adobe Creative Suite|
|Adobe After Effects CC||Version 188.8.131.524: Create Video which includes 3 Streams, 210 Frames, Render Multiple Frames Simultaneosly|
|Adobe Photoshop CC||Version 14.0 x64: Filter 15.7MB TIF Image: Radial Blur, Shape Blur, Median, Polar Coordinates|
|Adobe Premeire Pro CC||Version 7.0.0 (342), 6.61 GB MXF Project to H.264 to H.264 Blu-ray, Output 1920x1080, Maximum Quality|
|iTunes||Version 184.108.40.206 x64: Audio CD (Terminator II SE), 53 minutes, default AAC format|
|Lame MP3||Version 3.98.3: Audio CD "Terminator II SE", 53 min, convert WAV to MP3 audio format, Command: -b 160 --nores (160 kb/s)|
|Handbrake CLI||Version: 0.99: Video from Canon Eos 7D (1920x1080, 25 FPS) 1 Minutes 22 Seconds Audio: PCM-S16, 48000 Hz, 2-Channel, to Video: AVC1 Audio: AAC (High Profile)|
|TotalCodeStudio 2.5||Version: 220.127.116.1177: MPEG-2 to H.264, MainConcept H.264/AVC Codec, 28 sec HDTV 1920x1080 (MPEG-2), Audio: MPEG-2 (44.1 kHz, 2 Channel, 16-Bit, 224 kb/s), Codec: H.264 Pro, Mode: PAL 50i (25 FPS), Profile: H.264 BD HDMV|
|ABBYY FineReader||Version 10.0.102.95: Read PDF save to Doc, Source: Political Economy (J. Broadhurst 1842) 111 Pages|
|Adobe Acrobat 11||Version 18.104.22.1689: Print PDF from 115 Page PowerPoint, 128-bit RC4 Encryption|
|Autodesk 3ds Max 2013||Version 15.0 x64: Space Flyby Mentalray, 248 Frames, 1440x1080|
|Blender||Version: 2.68A, Cycles Engine, Syntax blender -b thg.blend -f 1, 1920x1080, 8x Anti-Aliasing, Render THG.blend frame 1|
|Visual Studio 2010||Version 10.0, Compile Google Chrome, Scripted|
|WinZip||Version 18.0 Pro: THG-Workload (1.3 GB) to ZIP, command line switches "-a -ez -p -r"|
|WinRAR||Version 5.0: THG-Workload (1.3 GB) to RAR, command line switches "winrar a -r -m3"|
|7-Zip||Version 9.30 alpha (64-bit): THG-Workload (1.3 GB) to .7z, command line switches "a -t7z -r -m0=LZMA2 -mx=5"|
|Synthetic Benchmarks and Settings|
|3DMark 11||Version: 22.214.171.124, Benchmark Only|
|3DMark Professional||Version: 126.96.36.199 (64-bit), Fire Strike Benchmark|
|PCMark 8||Version: 1.0.0 x64, Full Test|
|SiSoftware Sandra||Version 2014.02.20.10, CPU Test = CPU Arithmetic / Multimedia / Cryptography, Memory Bandwidth Benchmarks|
From the outset, we can see that DDR3-1600 C11 defaults seriously hinder the system's Physics score in 3DMark 11. Pushing the memory to DDR3-2400 allows it to nearly match the DDR4-2666, and the DDR4 latency penalty appears inconsequential.
Sandra's Cryptographic Bandwidth test brutally punishes the DDR3-1600 C11 configuration. DDR4-2666 has slightly more bandwidth than DDR3-2400, and the DDR4 latency penalty is only a minor factor in this test. Let's see how it all works in some real-world applications!
3D Game Benchmarks
Battlefield 4 has a 200 FPS limit, which also limits the impact of memory performance at any setting low enough to show the difference. Grid 2 on the other hand favors both low latency and high bandwidth, putting DDR3-2400 C11 on roughly equal footing with DDR4-2666 C15.
Arma 3 shows a slight preference for the latency advantage of DDR3-2400 C11, compared to DDR4-2666 C15. Far Cry 3 penalizes DDR3-1600, but it's hard to tell whether it's due to bandwidth or latency since the lower data rate corresponds to longer cycle time.
Adobe Premiere prefers the higher bandwidth of DDR4-2666, but most latency-affected apps put the performance of that memory on par with DDR3-2400.
Power, Heat And Efficiency
The DDR4 in this comparison had LEDs, while the DDR3 did not. LEDs don't consume more power when the memory is loaded though, and that's where the DDR4 board falls behind in spite of the fact that it was set to 1.30V, compared to DDR3 at 1.50V. From these tests, it appears that energy misers would be better off with low-voltage DDR3 than DDR4 of a similar voltage setting.
Thicker heat spreaders on the DDR3 modules required mounting the fan a little higher on the CPU cooler, thus increasing the temperature of the identical CPU voltage regulator of the DDR3 motherboard. Higher temperatures usually cause lower component efficiency, but the DDR3 configuration still pulled less overall energy to produce a greater efficiency rating.
|Frequency And Voltage settings|
|ASRock Fatal1ty Z170 Gaming K4/D3||ASRock Fatal1ty Z170 Gaming K4|
|BIOS||P1.50 (11/11/2015)||P2.20 (11/11/2015)|
|Base Clock||100-340 MHz (100 kHz)||100-340 MHz (100 kHz)|
|CPU Multiplier||8x-120x (1x)||8x-120x (1x)|
|DRAM Data Rates||800-4133 (100/133.3 MHz)||800-4133 (100/133.3 MHz)|
|CPU Vcore||0.90-1.52V (5 mV)||0.90-1.52V (5 mV)|
|System Agent||0.95-1.35V (10 mV)||0.95-1.35V (10 mV)|
|CPU I/O||0.85-1.25V (5 mV)||0.85-1.25V (5 mV)|
|PCH Voltage||0.90-1.30V (5 mV)||0.90-1.30V (5 mV)|
|DRAM Voltage||1.30-1.65V (5 mV)||1.00-1.80V (5 mV)|
|CAS Latency||4-31 Cycles||4-31 Cycles|
|tRCD||8-31 Cycles||8-31 Cycles|
|tRP||8-31 Cycles||8-31 Cycles|
|tRAS||28-63 Cycles||28-63 Cycles|
Remembering that this DDR3-vs-DDR4 comparison is also a comparison of otherwise-matching motherboards, it's good to see that both version have the same CPU overclocking capability. Unfortunately, DDR3 memory overclocking came up short, and even increased latency values couldn't push stability beyond DDR3-2533. Illustrated more clearly in the DDR4 numbers, memory overclocking stability appears to be limited by the output ceiling of each motherboard's DRAM voltage regulator.
While the price of the DDR4-bearing Fatal1ty Z170 Gaming K4 has recently increased to $135, the Fatal1ty Z170 Gaming K4/D3 is still available for $130. That hands the value win to DDR3, at least for now.
Onboard features and layout advantages were thoroughly discussed in the Fatal1ty Z170 Gaming K4 review. What else can be said of the Fatal1ty Z170 Gaming K4/D3?
Intel insists that Skylake is meant only for use with DDR4 and low-voltage DDR3, but motherboard manufacturers have found standard DDR3 perfectly compatible. While nobody is exactly sure yet how long the CPU will live at higher DDR3-appropriate memory voltage, overclockers have been pushing the CPU core to 1.45V for quite a while. Though the memory controller's slightly higher base voltage should allow it to support even higher signal voltage, we will recommend keeping the memory under 1.55V until we've seen adequate proof of appropriateness for a higher stopping point.